[ { "task": "poster detection", "subtask": "", "name": "ffb150e948e146c8badb30bb44723489.png", "path": "poster_ocr_1024/ffb150e948e146c8badb30bb44723489.png", "path_original": "poster_ocr/ffb150e948e146c8badb30bb44723489.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['REVOLUTION', 'Rock', 'Show', 'FRIDAY, 28 FEB.', 'SOUND OF', '@ THE INN', 'PERFORMERS', 'DOORS OPEN AT 9PM', 'MAIN STREET YOUR CITY', 'LIVE', 'WWW.WEBSITE.COM 555 666 444', 'SLAYMOORE', 'ROBSHOTS', 'THE OWLZ', '30$', 'ZEPPELIN', 'PRESENTS', '& SPECIAL GUEST', 'TICKETS', 'FACEBOOK', 'YOUTUBE', 'TWITTER', 'VIMEO']", "size": [ 2700, 3450 ], "texts": [ "REVOLUTION", "Rock", "Show", "FRIDAY, 28 FEB.", "SOUND OF", "@ THE INN", "PERFORMERS", "DOORS OPEN AT 9PM", "MAIN STREET YOUR CITY", "LIVE", "WWW.WEBSITE.COM 555 666 444", "SLAYMOORE", "ROBSHOTS", "THE OWLZ", "30$", "ZEPPELIN", "PRESENTS", "& SPECIAL GUEST", "TICKETS", "FACEBOOK", "YOUTUBE", "TWITTER", "VIMEO" ], "text_bbox": [ [ 432, 729, 1842, 1026 ], [ 514, 1245, 1399, 1689 ], [ 575, 1562, 1363, 1910 ], [ 1535, 1164, 2248, 1542 ], [ 509, 577, 1540, 822 ], [ 1591, 1282, 2217, 1663 ], [ 1461, 1885, 2309, 2055 ], [ 731, 2735, 1969, 2827 ], [ 729, 2632, 1977, 2706 ], [ 1556, 1724, 1983, 1921 ], [ 727, 2889, 1968, 2947 ], [ 1163, 2104, 1865, 2187 ], [ 1168, 2207, 1762, 2290 ], [ 1170, 2312, 1744, 2395 ], [ 391, 2243, 655, 2387 ], [ 1115, 269, 1592, 345 ], [ 1116, 357, 1599, 429 ], [ 1166, 2416, 1744, 2467 ], [ 398, 2391, 653, 2440 ], [ 617, 3174, 903, 3215 ], [ 1535, 3175, 1790, 3216 ], [ 1117, 3173, 1342, 3214 ], [ 1982, 3174, 2150, 3215 ] ], "bbox_areas": [ 418770, 392940, 274224, 269514, 252595, 238506, 144160, 113896, 92352, 84119, 71978, 58266, 49302, 47642, 38016, 36252, 34776, 29478, 12495, 11726, 10455, 9225, 6888 ] }, { "task": "poster detection", "subtask": "", "name": "ff28c53792f34523bceb25289f5665d1.png", "path": "poster_ocr_1024/ff28c53792f34523bceb25289f5665d1.png", "path_original": "poster_ocr/ff28c53792f34523bceb25289f5665d1.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['theweekender', '6 exterior pockets. Top zip and logo-lined, 3-pocket interior. Metal feet. Slim top handles.', 'www.yourblog.com']", "size": [ 750, 1050 ], "texts": [ "theweekender", "6 exterior pockets. Top zip and logo-lined, 3-pocket interior. Metal feet. Slim top handles.", "www.yourblog.com" ], "text_bbox": [ [ 76, 312, 676, 541 ], [ 98, 587, 656, 736 ], [ 204, 1002, 545, 1024 ] ], "bbox_areas": [ 137400, 83142, 7502 ] }, { "task": "poster detection", "subtask": "", "name": "fe531410b30641cba31ee0b89827ad3e.png", "path": "poster_ocr_1024/fe531410b30641cba31ee0b89827ad3e.png", "path_original": "poster_ocr/fe531410b30641cba31ee0b89827ad3e.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['e l l e', 'Adrianalima', 'cinque stilisthcona conteorie spericolate.edqnistiche es ecoglam', 'sobre las curvas,el amorcon una estrella delbasket y vivir en mabrld', 'bella &excitante', 'interviste-colpodi fulmine', 'neiau artieridark-chic conil.nuovo redelthriller', 'GETDRESSEDUP!', 'THESEFROCKSREALLYROCK', 'The digital training teacher zeng Jie', 'DIZIONARLOAMOROSO', 'londra', 'lui e lei digitali', '2013', '(en porlada)']", "size": [ 3744, 5616 ], "texts": [ "e l l e", "Adrianalima", "cinque stilisthcona conteorie spericolate.edqnistiche es ecoglam", "sobre las curvas,el amorcon una estrella delbasket y vivir en mabrld", "bella &excitante", "interviste-colpodi fulmine", "neiau artieridark-chic conil.nuovo redelthriller", "GETDRESSEDUP!", "THESEFROCKSREALLYROCK", "The digital training teacher zeng Jie", "DIZIONARLOAMOROSO", "londra", "lui e lei digitali", "2013", "(en porlada)" ], "text_bbox": [ [ 148, 163, 3623, 1510 ], [ 2081, 3388, 3651, 4151 ], [ 173, 2848, 1297, 3286 ], [ 2484, 4264, 3642, 4604 ], [ 2781, 2937, 3627, 3346 ], [ 172, 2465, 1200, 2801 ], [ 140, 3574, 832, 3996 ], [ 2597, 242, 3329, 634 ], [ 2588, 653, 3029, 1037 ], [ 174, 5275, 1244, 5425 ], [ 114, 4225, 804, 4407 ], [ 160, 3375, 698, 3537 ], [ 142, 4076, 798, 4205 ], [ 77, 4430, 485, 4614 ], [ 3017, 4171, 3623, 4261 ] ], "bbox_areas": [ 4680825, 1197910, 492312, 393720, 346014, 345408, 292024, 286944, 169344, 160500, 125580, 87156, 84624, 75072, 54540 ] }, { "task": "poster detection", "subtask": "", "name": "fdee5fc0d5e54071beb4f33c77df5337.png", "path": "poster_ocr_1024/fdee5fc0d5e54071beb4f33c77df5337.png", "path_original": "poster_ocr/fdee5fc0d5e54071beb4f33c77df5337.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['VIBES', 'VIBES', 'DOORS OPEN AT 9PM / 21+ TO ENTER MAIN STREET CITY CA. >>> ', '@ CLUB ZEPPELIN', 'MORE TEXT HERE OR DETAILS IF NECESSSARY. SPONOSORS, DETAILS ETC. DELETE THE TEXT AND USE YOUR LOGO HERE. MRE TEXT OR DETAILS IF NECESSARY, SPONSORS, INFO.', 'FEATURINGSPECIAL GUEST>>>', '35$', '21St', 'WWW.WEBSITE.COM ', 'ZEPPELIN PRESENTS', 'JUNE', 'DJ', 'tICKETS', 'SAtURDAY', '22ND EDITION']", "size": [ 2700, 3450 ], "texts": [ "VIBES", "VIBES", "DOORS OPEN AT 9PM / 21+ TO ENTER MAIN STREET CITY CA. >>> ", "@ CLUB ZEPPELIN", "MORE TEXT HERE OR DETAILS IF NECESSSARY. SPONOSORS, DETAILS ETC. DELETE THE TEXT AND USE YOUR LOGO HERE. MRE TEXT OR DETAILS IF NECESSARY, SPONSORS, INFO.", "FEATURINGSPECIAL GUEST>>>", "35$", "21St", "WWW.WEBSITE.COM ", "ZEPPELIN PRESENTS", "JUNE", "DJ", "tICKETS", "SAtURDAY", "22ND EDITION" ], "text_bbox": [ [ 171, 1763, 2520, 2483 ], [ 171, 1763, 2520, 2483 ], [ 179, 2773, 1683, 2975 ], [ 172, 2613, 1549, 2747 ], [ 179, 3054, 1867, 3125 ], [ 170, 1275, 543, 1491 ], [ 2051, 2736, 2430, 2932 ], [ 266, 590, 635, 728 ], [ 179, 3210, 1151, 3262 ], [ 2458, 333, 2508, 1311 ], [ 267, 469, 635, 581 ], [ 171, 1558, 397, 1715 ], [ 2050, 2939, 2435, 3024 ], [ 267, 399, 635, 458 ], [ 2403, 334, 2433, 882 ] ], "bbox_areas": [ 1691280, 1691280, 303808, 184518, 119848, 80568, 74284, 50922, 50544, 48900, 41216, 35482, 32725, 21712, 16440 ] }, { "task": "poster detection", "subtask": "", "name": "fdea6652aa8142b0a654526c06ff7cca.png", "path": "poster_ocr_1024/fdea6652aa8142b0a654526c06ff7cca.png", "path_original": "poster_ocr/fdea6652aa8142b0a654526c06ff7cca.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['10 best paintings of contemporary artists and illustrators', 'Art Gallery', '# New Article', 'www.creatogallery.com', 'Show Gallery']", "size": [ 1200, 1200 ], "texts": [ "10 best paintings of contemporary artists and illustrators", "Art Gallery", "# New Article", "www.creatogallery.com", "Show Gallery" ], "text_bbox": [ [ 231, 813, 972, 899 ], [ 249, 151, 766, 238 ], [ 852, 80, 999, 113 ], [ 780, 1048, 998, 1066 ], [ 257, 1048, 381, 1066 ] ], "bbox_areas": [ 63726, 44979, 4851, 3924, 2232 ] }, { "task": "poster detection", "subtask": "", "name": "fd864d73a032471698335a6289929f33.png", "path": "poster_ocr_1024/fd864d73a032471698335a6289929f33.png", "path_original": "poster_ocr/fd864d73a032471698335a6289929f33.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['ABSTRAKT', 'ABSTRAKT', 'dj PURE | Dj EDM | DJ MINIMAL', '9 PM . ENTRY 20$ . DRESS CODE : FUTURISTIC FANTASY', '-OVNI EVENT PRESENTs-', 'FRIDAY 15th JULY', 'facebook.com/YOURNAME', 'TWITTER.COM/YOURNAME', 'WWWW.WEBSITE.COM ']", "size": [ 1275, 1875 ], "texts": [ "ABSTRAKT", "ABSTRAKT", "dj PURE | Dj EDM | DJ MINIMAL", "9 PM . ENTRY 20$ . DRESS CODE : FUTURISTIC FANTASY", "-OVNI EVENT PRESENTs-", "FRIDAY 15th JULY", "facebook.com/YOURNAME", "TWITTER.COM/YOURNAME", "WWWW.WEBSITE.COM " ], "text_bbox": [ [ 123, 1280, 1152, 1464 ], [ 123, 1291, 1127, 1470 ], [ 215, 1573, 1061, 1615 ], [ 226, 1638, 1051, 1663 ], [ 258, 1234, 986, 1260 ], [ 511, 1499, 770, 1534 ], [ 264, 1667, 516, 1692 ], [ 795, 1667, 1030, 1692 ], [ 555, 1669, 757, 1691 ] ], "bbox_areas": [ 189336, 179716, 35532, 20625, 18928, 9065, 6300, 5875, 4444 ] }, { "task": "poster detection", "subtask": "", "name": "fd8612535bbf43c3bf39f9c705bb457b.png", "path": "poster_ocr_1024/fd8612535bbf43c3bf39f9c705bb457b.png", "path_original": "poster_ocr/fd8612535bbf43c3bf39f9c705bb457b.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['Slim', 'DOWN']", "size": [ 1080, 1920 ], "texts": [ "Slim", "DOWN" ], "text_bbox": [ [ 0, 24, 1054, 1033 ], [ 536, 772, 974, 850 ] ], "bbox_areas": [ 1063486, 34164 ] }, { "task": "poster detection", "subtask": "", "name": "fcb847641b074f2c875aef223098bf37.png", "path": "poster_ocr_1024/fcb847641b074f2c875aef223098bf37.png", "path_original": "poster_ocr/fcb847641b074f2c875aef223098bf37.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['HOUSE', 'DIGITAL MOTION PICTURES PRESENTS THE BEST DESIGN ARTINSPIRATIONS and VISIONS reserved 2015 COPYRIGHTS', 'THE DARK', 'KYLE HURT / JULIA LOPEZ / ADAM STANK', '-YOU ARE NOT ALONE-', 'WEBSITE COMING SOON', 'F WWW.YOURFACEBOOK.COM', '18.9.15', 'T www.YOURTWITTER.com', 'YOURPLACE']", "size": [ 1275, 1875 ], "texts": [ "HOUSE", "DIGITAL MOTION PICTURES PRESENTS THE BEST DESIGN ARTINSPIRATIONS and VISIONS reserved 2015 COPYRIGHTS", "THE DARK", "KYLE HURT / JULIA LOPEZ / ADAM STANK", "-YOU ARE NOT ALONE-", "WEBSITE COMING SOON", "F WWW.YOURFACEBOOK.COM", "18.9.15", "T www.YOURTWITTER.com", "YOURPLACE" ], "text_bbox": [ [ 260, 1053, 1013, 1202 ], [ 273, 1463, 1011, 1557 ], [ 278, 944, 993, 1039 ], [ 271, 1395, 1000, 1433 ], [ 321, 1211, 949, 1239 ], [ 303, 1753, 945, 1774 ], [ 737, 1658, 1026, 1686 ], [ 540, 1308, 746, 1347 ], [ 256, 1658, 533, 1686 ], [ 544, 1657, 728, 1694 ] ], "bbox_areas": [ 112197, 69372, 67925, 27702, 17584, 13482, 8092, 8034, 7756, 6808 ] }, { "task": "poster detection", "subtask": "", "name": "fc081dac97bc43ffbc26c91737a96d0d.png", "path": "poster_ocr_1024/fc081dac97bc43ffbc26c91737a96d0d.png", "path_original": "poster_ocr/fc081dac97bc43ffbc26c91737a96d0d.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['COLOR', 'SAT9 DEC/ 20:00', 'SOUND', 'LOPOSTACAG', 'KIKAZ ARENA', 'MASTAYAKA', 'WWW.LIPASDA.COM', 'SISIKALIO', 'Of', 'COTRA PRESENT', 'TELP. 9282 298', '$29', 'FEATURED', 'ENTRY']", "size": [ 1275, 1875 ], "texts": [ "COLOR", "SAT9 DEC/ 20:00", "SOUND", "LOPOSTACAG", "KIKAZ ARENA", "MASTAYAKA", "WWW.LIPASDA.COM", "SISIKALIO", "Of", "COTRA PRESENT", "TELP. 9282 298", "$29", "FEATURED", "ENTRY" ], "text_bbox": [ [ 490, 903, 803, 974 ], [ 110, 96, 498, 136 ], [ 571, 988, 807, 1041 ], [ 924, 1609, 1189, 1641 ], [ 937, 1457, 1190, 1489 ], [ 969, 1649, 1190, 1681 ], [ 897, 1739, 1189, 1763 ], [ 987, 1567, 1190, 1599 ], [ 475, 984, 559, 1052 ], [ 964, 1423, 1190, 1447 ], [ 979, 1706, 1190, 1729 ], [ 232, 1308, 323, 1354 ], [ 1056, 1530, 1189, 1554 ], [ 228, 1365, 322, 1390 ] ], "bbox_areas": [ 22223, 15520, 12508, 8480, 8096, 7072, 7008, 6496, 5712, 5424, 4853, 4186, 3192, 2350 ] }, { "task": "poster detection", "subtask": "", "name": "fb31c6bd58f640f28404a860e0c75a66.png", "path": "poster_ocr_1024/fb31c6bd58f640f28404a860e0c75a66.png", "path_original": "poster_ocr/fb31c6bd58f640f28404a860e0c75a66.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['nature', 'look deep into', 'everything better', 'and you will understand']", "size": [ 3342, 3338 ], "texts": [ "nature", "look deep into", "everything better", "and you will understand" ], "text_bbox": [ [ 1114, 1215, 2245, 1756 ], [ 1122, 900, 2251, 1141 ], [ 1110, 2045, 2237, 2235 ], [ 1109, 1828, 2227, 1967 ] ], "bbox_areas": [ 611871, 272089, 214130, 155402 ] }, { "task": "poster detection", "subtask": "", "name": "fada943865064ad1ba2279cefa30eafe.png", "path": "poster_ocr_1024/fada943865064ad1ba2279cefa30eafe.png", "path_original": "poster_ocr/fada943865064ad1ba2279cefa30eafe.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['New Video', 'Recipes of Amanda Florence', 'Amanda YouTube Channel', 'www.creatostories.com']", "size": [ 1200, 1200 ], "texts": [ "New Video", "Recipes of Amanda Florence", "Amanda YouTube Channel", "www.creatostories.com" ], "text_bbox": [ [ 204, 500, 993, 614 ], [ 443, 89, 754, 136 ], [ 410, 262, 853, 286 ], [ 921, 1124, 1139, 1139 ] ], "bbox_areas": [ 89946, 14617, 10632, 3270 ] }, { "task": "poster detection", "subtask": "", "name": "fa9bb1afe56b436db7b1e32256385ade.png", "path": "poster_ocr_1024/fa9bb1afe56b436db7b1e32256385ade.png", "path_original": "poster_ocr/fa9bb1afe56b436db7b1e32256385ade.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['MUSICNIGHTS', 'MUSICNIGHTS', 'MUSICNIGHTS', '$99Shotphon', '$99Shotphon', '$99Shotphon', 'Dedicate', 'Dedicate', 'Dedicate', 'Our mission is to controll and access all the ways thatcan build up success road', 'Our mission is to controll and access all the ways thatcan build up success road', 'Our mission is to controll and access all the ways thatcan build up success road', 'Our mission is to controll and access all the ways thatcan build up success road', 'Our mission is to controll and access all the ways thatcan build up success road', 'Our mission is to controll and access all the ways thatcan build up success road', 'MondaySpecialwith Silvacoda', 'MondaySpecialwith Silvacoda', 'MondaySpecialwith Silvacoda', 'Music Clubs', 'Music Clubs', 'Music Clubs', 'Promotion', 'Promotion', 'Promotion', '25 July', '25 July', '25 July', '2015', '2015', '2015', 'www.yourstudio.com', 'www.yourstudio.com', 'www.yourstudio.com', 'www.yourstudio.com', 'www.yourstudio.com', 'www.yourstudio.com']", "size": [ 3239, 4763 ], "texts": [ "MUSICNIGHTS", "MUSICNIGHTS", "MUSICNIGHTS", "$99Shotphon", "$99Shotphon", "$99Shotphon", "Dedicate", "Dedicate", "Dedicate", "Our mission is to controll and access all the ways thatcan build up success road", "Our mission is to controll and access all the ways thatcan build up success road", "Our mission is to controll and access all the ways thatcan build up success road", "Our mission is to controll and access all the ways thatcan build up success road", "Our mission is to controll and access all the ways thatcan build up success road", "Our mission is to controll and access all the ways thatcan build up success road", "MondaySpecialwith Silvacoda", "MondaySpecialwith Silvacoda", "MondaySpecialwith Silvacoda", "Music Clubs", "Music Clubs", "Music Clubs", "Promotion", "Promotion", "Promotion", "25 July", "25 July", "25 July", "2015", "2015", "2015", "www.yourstudio.com", "www.yourstudio.com", "www.yourstudio.com", "www.yourstudio.com", "www.yourstudio.com", "www.yourstudio.com" ], "text_bbox": [ [ 728, 1958, 2519, 2770 ], [ 728, 1958, 2519, 2770 ], [ 728, 1958, 2519, 2770 ], [ 1802, 307, 2987, 829 ], [ 1802, 307, 2987, 829 ], [ 1802, 307, 2987, 829 ], [ 928, 2980, 2254, 3304 ], [ 928, 2980, 2254, 3304 ], [ 928, 2980, 2254, 3304 ], [ 982, 1672, 2264, 1856 ], [ 982, 1672, 2264, 1856 ], [ 982, 1672, 2264, 1856 ], [ 226, 4309, 1524, 4466 ], [ 226, 4309, 1524, 4466 ], [ 226, 4309, 1524, 4466 ], [ 488, 3918, 1108, 4184 ], [ 488, 3918, 1108, 4184 ], [ 488, 3918, 1108, 4184 ], [ 249, 368, 1173, 528 ], [ 249, 368, 1173, 528 ], [ 249, 368, 1173, 528 ], [ 1173, 1477, 2092, 1619 ], [ 1173, 1477, 2092, 1619 ], [ 1173, 1477, 2092, 1619 ], [ 439, 3640, 1042, 3806 ], [ 439, 3640, 1042, 3806 ], [ 439, 3640, 1042, 3806 ], [ 229, 3653, 412, 4184 ], [ 229, 3653, 412, 4184 ], [ 229, 3653, 412, 4184 ], [ 1153, 2859, 2111, 2955 ], [ 1153, 2859, 2111, 2955 ], [ 1153, 2859, 2111, 2955 ], [ 224, 4489, 1023, 4567 ], [ 224, 4489, 1023, 4567 ], [ 224, 4489, 1023, 4567 ] ], "bbox_areas": [ 1454292, 1454292, 1454292, 618570, 618570, 618570, 429624, 429624, 429624, 235888, 235888, 235888, 203786, 203786, 203786, 164920, 164920, 164920, 147840, 147840, 147840, 130498, 130498, 130498, 100098, 100098, 100098, 97173, 97173, 97173, 91968, 91968, 91968, 62322, 62322, 62322 ] }, { "task": "poster detection", "subtask": "", "name": "f7d5fbfe8af4463ba86e54ea42be0839.png", "path": "poster_ocr_1024/f7d5fbfe8af4463ba86e54ea42be0839.png", "path_original": "poster_ocr/f7d5fbfe8af4463ba86e54ea42be0839.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['new collection', 'Urban creative inspiration glitter.', 'Peace and Love', 'www.domainaddress.com']", "size": [ 1200, 1200 ], "texts": [ "new collection", "Urban creative inspiration glitter.", "Peace and Love", "www.domainaddress.com" ], "text_bbox": [ [ 343, 186, 856, 227 ], [ 322, 978, 878, 1011 ], [ 478, 1076, 722, 1112 ], [ 454, 72, 748, 90 ] ], "bbox_areas": [ 21033, 18348, 8784, 5292 ] }, { "task": "poster detection", "subtask": "", "name": "f7464464487a4d4e8add7e5c5e8219c0.png", "path": "poster_ocr_1024/f7464464487a4d4e8add7e5c5e8219c0.png", "path_original": "poster_ocr/f7464464487a4d4e8add7e5c5e8219c0.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['LOVA', 'Frios', '7. 12. 2015', 'heroes -las vegas resorts - casino', 'END OF SUMMER BEATS', 'flyerheroes.com presents']", "size": [ 1275, 1875 ], "texts": [ "LOVA", "Frios", "7. 12. 2015", "heroes -las vegas resorts - casino", "END OF SUMMER BEATS", "flyerheroes.com presents" ], "text_bbox": [ [ 321, 562, 987, 766 ], [ 319, 773, 999, 965 ], [ 400, 1443, 911, 1532 ], [ 214, 1589, 1060, 1620 ], [ 341, 987, 973, 1016 ], [ 328, 525, 908, 546 ] ], "bbox_areas": [ 135864, 130560, 45479, 26226, 18328, 12180 ] }, { "task": "poster detection", "subtask": "", "name": "f66d151300c64d4e880e56770b617635.png", "path": "poster_ocr_1024/f66d151300c64d4e880e56770b617635.png", "path_original": "poster_ocr/f66d151300c64d4e880e56770b617635.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['50', 'Pick your favorite productand use the code:', 'Exclusive Sale', '50% OFF', 'SPRING2017']", "size": [ 1080, 1080 ], "texts": [ "50", "Pick your favorite productand use the code:", "Exclusive Sale", "50% OFF", "SPRING2017" ], "text_bbox": [ [ 326, 554, 754, 846 ], [ 346, 880, 735, 966 ], [ 398, 38, 675, 73 ], [ 440, 687, 641, 718 ], [ 438, 1008, 642, 1026 ] ], "bbox_areas": [ 124976, 33454, 9695, 6231, 3672 ] }, { "task": "poster detection", "subtask": "", "name": "f658384fea2e47dd8f28f6dde934d7ab.png", "path": "poster_ocr_1024/f658384fea2e47dd8f28f6dde934d7ab.png", "path_original": "poster_ocr/f658384fea2e47dd8f28f6dde934d7ab.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['NEONPARTY ', 'YOUR ENTERTAINMENTPRESENTS', 'ADMISSION 10$ | FREE FOR LADIES | START 9:30 PM', 'Dj FASHIONgroove', 'Dj ELEKTROELECTRO', 'Nightclub', 'Dj NEONHip-hop', '15 | 10 | 14', 'F facebook.com/YOURNAME', 'T twitter.com/YOURNAME', 'W wWw.website.com', 'www.clubname.com']", "size": [ 1275, 1875 ], "texts": [ "NEONPARTY ", "YOUR ENTERTAINMENTPRESENTS", "ADMISSION 10$ | FREE FOR LADIES | START 9:30 PM", "Dj FASHIONgroove", "Dj ELEKTROELECTRO", "Nightclub", "Dj NEONHip-hop", "15 | 10 | 14", "F facebook.com/YOURNAME", "T twitter.com/YOURNAME", "W wWw.website.com", "www.clubname.com" ], "text_bbox": [ [ 129, 604, 1150, 1249 ], [ 424, 268, 856, 369 ], [ 200, 1532, 1092, 1569 ], [ 536, 1346, 760, 1447 ], [ 857, 1346, 1075, 1447 ], [ 516, 1660, 772, 1743 ], [ 242, 1344, 401, 1445 ], [ 545, 159, 735, 233 ], [ 199, 1570, 494, 1601 ], [ 544, 1570, 815, 1601 ], [ 862, 1570, 1092, 1600 ], [ 529, 1739, 751, 1765 ] ], "bbox_areas": [ 658545, 43632, 33004, 22624, 22018, 21248, 16059, 14060, 9145, 8401, 6900, 5772 ] }, { "task": "poster detection", "subtask": "", "name": "f61bfaaaefbf4fea998c5f0049cdd403.png", "path": "poster_ocr_1024/f61bfaaaefbf4fea998c5f0049cdd403.png", "path_original": "poster_ocr/f61bfaaaefbf4fea998c5f0049cdd403.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['CLUB', 'CAMALL', 'CLUBDOORS OPEN AT 10PM3HOUR PREMIUM', 'MORE INFO CALL: 12345678OPEN CLUB 10PM']", "size": [ 1583, 2183 ], "texts": [ "CLUB", "CAMALL", "CLUBDOORS OPEN AT 10PM3HOUR PREMIUM", "MORE INFO CALL: 12345678OPEN CLUB 10PM" ], "text_bbox": [ [ 396, 958, 1151, 1180 ], [ 363, 1874, 1189, 2053 ], [ 224, 1746, 710, 1829 ], [ 867, 1748, 1345, 1832 ] ], "bbox_areas": [ 167610, 147854, 40338, 40152 ] }, { "task": "poster detection", "subtask": "", "name": "f54a151b43174444b88a5bd1ede1a5bb.png", "path": "poster_ocr_1024/f54a151b43174444b88a5bd1ede1a5bb.png", "path_original": "poster_ocr/f54a151b43174444b88a5bd1ede1a5bb.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['About', '/Me/', 'Jason Wolf Creative Designer', '@ CreatoInstagram']", "size": [ 1200, 1200 ], "texts": [ "About", "/Me/", "Jason Wolf Creative Designer", "@ CreatoInstagram" ], "text_bbox": [ [ 277, 207, 922, 356 ], [ 354, 789, 844, 977 ], [ 409, 991, 788, 1045 ], [ 129, 890, 147, 1070 ] ], "bbox_areas": [ 96105, 92120, 20466, 3240 ] }, { "task": "poster detection", "subtask": "", "name": "f495f1e087dc44edbc0607885a09725a.png", "path": "poster_ocr_1024/f495f1e087dc44edbc0607885a09725a.png", "path_original": "poster_ocr/f495f1e087dc44edbc0607885a09725a.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['URBAN', 'URBAN', 'Jungle', 'SATURDAY JUNE 21ST', 'DOORS OPEN AT 9PM // FREE PARKING // 1 FREE COCKTAIL AT THE BAR', '5422 MAIN STREET, YOUR CITY CA. NEAR THE PLAZA', '@CLUBZEPPELIN', '25$', 'ENTRY', 'FOR MORE INFO WWW.WEBSITE.COM OR CAL US 6544 433 333', 'FACEBOOK/CLUBNAME', 'TWITTER/CLUBNAME']", "size": [ 2700, 3450 ], "texts": [ "URBAN", "URBAN", "Jungle", "SATURDAY JUNE 21ST", "DOORS OPEN AT 9PM // FREE PARKING // 1 FREE COCKTAIL AT THE BAR", "5422 MAIN STREET, YOUR CITY CA. NEAR THE PLAZA", "@CLUBZEPPELIN", "25$", "ENTRY", "FOR MORE INFO WWW.WEBSITE.COM OR CAL US 6544 433 333", "FACEBOOK/CLUBNAME", "TWITTER/CLUBNAME" ], "text_bbox": [ [ 238, 1403, 2453, 2399 ], [ 301, 1400, 2389, 2401 ], [ 1236, 1943, 2393, 2756 ], [ 302, 2472, 1571, 2655 ], [ 370, 2850, 2332, 2945 ], [ 538, 2959, 2167, 3021 ], [ 340, 2672, 1124, 2762 ], [ 292, 165, 558, 416 ], [ 295, 398, 558, 546 ], [ 778, 3050, 1923, 3081 ], [ 1422, 3244, 1847, 3271 ], [ 1966, 3243, 2353, 3270 ] ], "bbox_areas": [ 2206140, 2090088, 940641, 232227, 186390, 100998, 70560, 66766, 38924, 35495, 11475, 10449 ] }, { "task": "poster detection", "subtask": "", "name": "f45d3c7eb4eb41c8978e70a60c77448c.png", "path": "poster_ocr_1024/f45d3c7eb4eb41c8978e70a60c77448c.png", "path_original": "poster_ocr/f45d3c7eb4eb41c8978e70a60c77448c.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['HelloSpring']", "size": [ 3334, 3333 ], "texts": [ "HelloSpring" ], "text_bbox": [ [ 920, 280, 2386, 1579 ] ], "bbox_areas": [ 1904334 ] }, { "task": "poster detection", "subtask": "", "name": "f4339b8ee7ee474b93dbdf4aa7d3123c.png", "path": "poster_ocr_1024/f4339b8ee7ee474b93dbdf4aa7d3123c.png", "path_original": "poster_ocr/f4339b8ee7ee474b93dbdf4aa7d3123c.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['beautiful', 'kids', 'Childhood swing', 'Sunny weather the children enjoy sunshine and happiness']", "size": [ 4000, 6000 ], "texts": [ "beautiful", "kids", "Childhood swing", "Sunny weather the children enjoy sunshine and happiness" ], "text_bbox": [ [ 286, 501, 3541, 2410 ], [ 797, 403, 3180, 1203 ], [ 908, 1270, 3056, 1440 ], [ 705, 5710, 3342, 5776 ] ], "bbox_areas": [ 6213795, 1906400, 365160, 174042 ] }, { "task": "poster detection", "subtask": "", "name": "f10a8e2e646149cab39524439bb3986c.png", "path": "poster_ocr_1024/f10a8e2e646149cab39524439bb3986c.png", "path_original": "poster_ocr/f10a8e2e646149cab39524439bb3986c.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['man ', 'with a beard longer than 30 cm free', 'party', 'NameClub', 'PARTY DRIVE MIAMI BEACH FL 33139 I 016-101-1985, WWW.NIGHTCLUB.COM I VIPTABLES@NIGHTCLUB.COM', ' bearded ', 'FREE ENTRY BEFORE 11 PM I GIRLS FREE ENTRY, DOORS OPEN AT 9PM I AFTER 11 PM ADMISSION $12', 'dj bear', 'special guest', 'styleflyers present']", "size": [ 1350, 1950 ], "texts": [ "man ", "with a beard longer than 30 cm free", "party", "NameClub", "PARTY DRIVE MIAMI BEACH FL 33139 I 016-101-1985, WWW.NIGHTCLUB.COM I VIPTABLES@NIGHTCLUB.COM", " bearded ", "FREE ENTRY BEFORE 11 PM I GIRLS FREE ENTRY, DOORS OPEN AT 9PM I AFTER 11 PM ADMISSION $12", "dj bear", "special guest", "styleflyers present" ], "text_bbox": [ [ 442, 654, 906, 833 ], [ 224, 1322, 395, 1792 ], [ 420, 874, 932, 1008 ], [ 404, 327, 950, 440 ], [ 1205, 93, 1242, 1750 ], [ 410, 511, 975, 617 ], [ 65, 341, 104, 1859 ], [ 452, 1476, 906, 1576 ], [ 373, 1360, 994, 1428 ], [ 432, 77, 922, 149 ] ], "bbox_areas": [ 83056, 80370, 68608, 61698, 61309, 59890, 59202, 45400, 42228, 35280 ] }, { "task": "poster detection", "subtask": "", "name": "f0f968c0e84e4db5bb8353b55316b6e0.png", "path": "poster_ocr_1024/f0f968c0e84e4db5bb8353b55316b6e0.png", "path_original": "poster_ocr/f0f968c0e84e4db5bb8353b55316b6e0.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['ALTERNATIVE NATION', 'SATURDAY AUGUST 28TH', 'LIVE IN CONCERT AT ZEPPELIN CLUB', 'WWW.INDIEFEST.COM', '2 FREE DRINKS AT THE BAR // DOORS OPEN AT 9PM ', 'FACEBOOK/INDIEFEST TWITTER/INDIEFEST YOUTUBE/INDIEFEST', '25$', '6523 MAIN STREET YOUR CITY CA. NEAR THE PLAZA', 'THE ROOTS', 'BLACKMOORE', 'ZEPPELIN & RADIO 54 PRESENT', 'ALL NIGHT', 'BROUGHT TO YOU BY THE INN AND RADIO 54', 'ENTRY', 'CALL US AT:', 'FROM 1AM TO 3AM', '5554 445 445', 'FROM 10PM TO 1AM']", "size": [ 2700, 3450 ], "texts": [ "ALTERNATIVE NATION", "SATURDAY AUGUST 28TH", "LIVE IN CONCERT AT ZEPPELIN CLUB", "WWW.INDIEFEST.COM", "2 FREE DRINKS AT THE BAR // DOORS OPEN AT 9PM ", "FACEBOOK/INDIEFEST TWITTER/INDIEFEST YOUTUBE/INDIEFEST", "25$", "6523 MAIN STREET YOUR CITY CA. NEAR THE PLAZA", "THE ROOTS", "BLACKMOORE", "ZEPPELIN & RADIO 54 PRESENT", "ALL NIGHT", "BROUGHT TO YOU BY THE INN AND RADIO 54", "ENTRY", "CALL US AT:", "FROM 1AM TO 3AM", "5554 445 445", "FROM 10PM TO 1AM" ], "text_bbox": [ [ 272, 295, 2428, 608 ], [ 473, 2402, 2238, 2601 ], [ 270, 2193, 2428, 2296 ], [ 404, 2969, 1433, 3116 ], [ 405, 2775, 1812, 2864 ], [ 403, 3233, 2290, 3271 ], [ 1906, 2700, 2260, 2897 ], [ 403, 2871, 1814, 2915 ], [ 1945, 1955, 2411, 2076 ], [ 254, 782, 762, 888 ], [ 948, 174, 1864, 225 ], [ 1904, 2977, 2261, 3084 ], [ 404, 2686, 1391, 2721 ], [ 1905, 2882, 2261, 2968 ], [ 1467, 2977, 1802, 3065 ], [ 2000, 1905, 2408, 1942 ], [ 1466, 3074, 1800, 3114 ], [ 254, 736, 618, 769 ] ], "bbox_areas": [ 674828, 351235, 222274, 151263, 125223, 71706, 69738, 62084, 56386, 53848, 46716, 38199, 34545, 30616, 29480, 15096, 13360, 12012 ] }, { "task": "poster detection", "subtask": "", "name": "f049ec7b408844f79212d29ea0dc7f95.png", "path": "poster_ocr_1024/f049ec7b408844f79212d29ea0dc7f95.png", "path_original": "poster_ocr/f049ec7b408844f79212d29ea0dc7f95.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['BIGCITY', 'DJ LEX.MC SPARK.DJ EWON', '19/06/2016', 'ELECTRO // DUBSTEP // HIP HOP // FREESTYLES', 'www.yourwebsite.com', 'YOURPLACE']", "size": [ 1275, 1875 ], "texts": [ "BIGCITY", "DJ LEX.MC SPARK.DJ EWON", "19/06/2016", "ELECTRO // DUBSTEP // HIP HOP // FREESTYLES", "www.yourwebsite.com", "YOURPLACE" ], "text_bbox": [ [ 178, 152, 1127, 1139 ], [ 174, 1390, 1104, 1450 ], [ 355, 1251, 924, 1345 ], [ 169, 1467, 1106, 1502 ], [ 334, 1638, 941, 1668 ], [ 530, 1569, 744, 1614 ] ], "bbox_areas": [ 936663, 55800, 53486, 32795, 18210, 9630 ] }, { "task": "poster detection", "subtask": "", "name": "f017431f2d414c119b9a7022cb5903a1.png", "path": "poster_ocr_1024/f017431f2d414c119b9a7022cb5903a1.png", "path_original": "poster_ocr/f017431f2d414c119b9a7022cb5903a1.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['ELECTRO', 'TAKEOVER', '35$ ENTRY // FREE FOR GIRLS // VALET PARKING', 'SATURDAYJULY 14TH', '342 MAIN STREET, YOUR CITY, CA. NEAR THE PLAZA', '@CLUBZEPPEIN', 'DJ MAC DAVE', 'WWW.WEBSITE.COM', 'DOORS OPEN AT 9PM', '2014 SUMMERFEST', 'ZEPPELIN PRESENTS', 'DJ SLICK', 'DJ PAUL', '666 555 433', 'FROM 03 TO 06', 'FROM 12 TO 03', 'FROM 9 TO 12']", "size": [ 2700, 3450 ], "texts": [ "ELECTRO", "TAKEOVER", "35$ ENTRY // FREE FOR GIRLS // VALET PARKING", "SATURDAYJULY 14TH", "342 MAIN STREET, YOUR CITY, CA. NEAR THE PLAZA", "@CLUBZEPPEIN", "DJ MAC DAVE", "WWW.WEBSITE.COM", "DOORS OPEN AT 9PM", "2014 SUMMERFEST", "ZEPPELIN PRESENTS", "DJ SLICK", "DJ PAUL", "666 555 433", "FROM 03 TO 06", "FROM 12 TO 03", "FROM 9 TO 12" ], "text_bbox": [ [ 759, 1453, 1936, 1706 ], [ 766, 1750, 1933, 1974 ], [ 473, 2817, 2218, 2885 ], [ 838, 335, 1853, 412 ], [ 650, 2919, 2052, 2968 ], [ 1021, 2124, 1674, 2204 ], [ 1111, 485, 1585, 539 ], [ 1071, 3047, 1619, 3089 ], [ 1082, 2207, 1623, 2247 ], [ 1120, 1213, 1585, 1253 ], [ 1114, 156, 1584, 193 ], [ 475, 487, 780, 541 ], [ 1918, 487, 2220, 540 ], [ 1184, 3102, 1509, 3143 ], [ 1942, 564, 2193, 590 ], [ 1227, 562, 1466, 588 ], [ 509, 564, 728, 588 ] ], "bbox_areas": [ 297781, 261408, 118660, 78155, 68698, 52240, 25596, 23016, 21640, 18600, 17390, 16470, 16006, 13325, 6526, 6214, 5256 ] }, { "task": "poster detection", "subtask": "", "name": "efbf1f97e2f74de8b3c6345bb33b60d8.png", "path": "poster_ocr_1024/efbf1f97e2f74de8b3c6345bb33b60d8.png", "path_original": "poster_ocr/efbf1f97e2f74de8b3c6345bb33b60d8.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n[' Alatetur, adipisci velit, sed quia non numquam eius modi tempora quaerat voluptatem.', 'the classics', 'hat collection', 'shop']", "size": [ 750, 1050 ], "texts": [ " Alatetur, adipisci velit, sed quia non numquam eius modi tempora quaerat voluptatem.", "the classics", "hat collection", "shop" ], "text_bbox": [ [ 88, 757, 669, 835 ], [ 84, 539, 665, 585 ], [ 171, 638, 580, 666 ], [ 293, 891, 457, 926 ] ], "bbox_areas": [ 45318, 26726, 11452, 5740 ] }, { "task": "poster detection", "subtask": "", "name": "efbd5eb34a864d3e9bcf67ed79c63b09.png", "path": "poster_ocr_1024/efbd5eb34a864d3e9bcf67ed79c63b09.png", "path_original": "poster_ocr/efbd5eb34a864d3e9bcf67ed79c63b09.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['brewfest', 'Awesome beer, Cool music, & Great Food! Featuring dozens of international and domestic beers', 'free', 'OCTOBER 26TH', 'entrance', '123 MAIN STREET, NY, NY', 'because... beer', 'WEBPAGE.COM']", "size": [ 1275, 1875 ], "texts": [ "brewfest", "Awesome beer, Cool music, & Great Food! Featuring dozens of international and domestic beers", "free", "OCTOBER 26TH", "entrance", "123 MAIN STREET, NY, NY", "because... beer", "WEBPAGE.COM" ], "text_bbox": [ [ 179, 937, 1083, 1187 ], [ 424, 1293, 843, 1533 ], [ 873, 1457, 1107, 1665 ], [ 435, 802, 831, 875 ], [ 904, 1640, 1112, 1743 ], [ 172, 1656, 603, 1702 ], [ 479, 1206, 788, 1252 ], [ 264, 1739, 512, 1779 ] ], "bbox_areas": [ 226000, 100560, 48672, 28908, 21424, 19826, 14214, 9920 ] }, { "task": "poster detection", "subtask": "", "name": "ee9aca9951e548fca8a9c06403229ec7.png", "path": "poster_ocr_1024/ee9aca9951e548fca8a9c06403229ec7.png", "path_original": "poster_ocr/ee9aca9951e548fca8a9c06403229ec7.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['HIGH', 'MOOVE', '03/11/2015', 'dj FLYER | Dj LIQUID | DJ GRAPHICS', 'Club designers 123 St. YOUR CITY', 'ENTRY 10$ free admission for ladies until 10', '-YOUR ENTERTAINMENT PRESENTS-', 'F facebook.com/YOURNAME', 'T twitter.com/YOURNAME', 'W www.yourname.com']", "size": [ 1275, 1875 ], "texts": [ "HIGH", "MOOVE", "03/11/2015", "dj FLYER | Dj LIQUID | DJ GRAPHICS", "Club designers 123 St. YOUR CITY", "ENTRY 10$ free admission for ladies until 10", "-YOUR ENTERTAINMENT PRESENTS-", "F facebook.com/YOURNAME", "T twitter.com/YOURNAME", "W www.yourname.com" ], "text_bbox": [ [ 209, 645, 1067, 951 ], [ 209, 973, 1066, 1166 ], [ 204, 290, 1067, 419 ], [ 205, 1360, 1068, 1438 ], [ 207, 1444, 1066, 1501 ], [ 207, 1509, 1063, 1540 ], [ 272, 205, 1003, 236 ], [ 203, 1557, 498, 1586 ], [ 798, 1555, 1069, 1584 ], [ 519, 1555, 774, 1583 ] ], "bbox_areas": [ 262548, 165401, 111327, 67314, 48963, 26536, 22661, 8555, 7859, 7140 ] }, { "task": "poster detection", "subtask": "", "name": "edb67e5c3e5147a289dea017b47abaca.png", "path": "poster_ocr_1024/edb67e5c3e5147a289dea017b47abaca.png", "path_original": "poster_ocr/edb67e5c3e5147a289dea017b47abaca.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['ELECTRO', 'TAKEOVER', '1894 MAIN STREET, CITY, CA. NEAR THE PLAZA ', '20$ ENTRY FREE FOR GIRLS 5$ DRINKS VALET PARKING', '@ CLUB ZEPPELIN', 'DOORS OPEN AT 9PM', 'WWW.SUMMERFEST.COM', 'OR CALL US: 555 666 444', '2014', 'SATURDAY', 'SUMMERFEST', 'DJ SLICK D', 'DJ SHADE', 'JULY 14TH', 'DJ PAULY', 'ZEPPELIN PRESENTS', 'DJ VOX', 'EDITION', 'FACEBOOK', 'YOUTUBE', 'TWITTER', 'VIMEO']", "size": [ 2700, 3450 ], "texts": [ "ELECTRO", "TAKEOVER", "1894 MAIN STREET, CITY, CA. NEAR THE PLAZA ", "20$ ENTRY FREE FOR GIRLS 5$ DRINKS VALET PARKING", "@ CLUB ZEPPELIN", "DOORS OPEN AT 9PM", "WWW.SUMMERFEST.COM", "OR CALL US: 555 666 444", "2014", "SATURDAY", "SUMMERFEST", "DJ SLICK D", "DJ SHADE", "JULY 14TH", "DJ PAULY", "ZEPPELIN PRESENTS", "DJ VOX", "EDITION", "FACEBOOK", "YOUTUBE", "TWITTER", "VIMEO" ], "text_bbox": [ [ 857, 1523, 1841, 1738 ], [ 857, 1751, 1840, 1939 ], [ 498, 2779, 2204, 2849 ], [ 500, 2691, 2203, 2754 ], [ 929, 639, 1776, 726 ], [ 1033, 748, 1666, 797 ], [ 284, 3093, 961, 3136 ], [ 1764, 3095, 2417, 3138 ], [ 1214, 1253, 1484, 1353 ], [ 159, 1687, 565, 1750 ], [ 1120, 2100, 1575, 2156 ], [ 841, 330, 1242, 393 ], [ 1501, 326, 1888, 389 ], [ 2141, 1695, 2532, 1755 ], [ 284, 330, 659, 390 ], [ 1082, 160, 1619, 201 ], [ 2121, 327, 2415, 389 ], [ 1214, 1358, 1484, 1418 ], [ 366, 3254, 658, 3294 ], [ 2158, 3252, 2417, 3292 ], [ 1037, 3256, 1275, 3295 ], [ 1646, 3254, 1819, 3294 ] ], "bbox_areas": [ 211560, 184804, 119420, 107289, 73689, 31017, 29111, 28079, 27000, 25578, 25480, 25263, 24381, 23460, 22500, 22017, 18228, 16200, 11680, 10360, 9282, 6920 ] }, { "task": "poster detection", "subtask": "", "name": "ed30e604b7a142b385379ff92a53c42e.png", "path": "poster_ocr_1024/ed30e604b7a142b385379ff92a53c42e.png", "path_original": "poster_ocr/ed30e604b7a142b385379ff92a53c42e.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['NewDresses.', 'All of the latest styles are just a click away', 'shop new arrivals']", "size": [ 750, 1050 ], "texts": [ "NewDresses.", "All of the latest styles are just a click away", "shop new arrivals" ], "text_bbox": [ [ 102, 378, 633, 613 ], [ 103, 661, 629, 688 ], [ 135, 747, 446, 765 ] ], "bbox_areas": [ 124785, 14202, 5598 ] }, { "task": "poster detection", "subtask": "", "name": "ec5b439344a443748fde0378867ee33e.png", "path": "poster_ocr_1024/ec5b439344a443748fde0378867ee33e.png", "path_original": "poster_ocr/ec5b439344a443748fde0378867ee33e.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['SUmmerMUsic', 'BUSINESSINSPIRN DEATINS', '18Jan', 'BUSINESSINSPIRATION Adds', 'To construcon project is coand otheron project is colentcts ', 'Call us :+098 123 4556 234', 'Business Designs', 'DJ Low-J & Yolk-X']", "size": [ 1275, 1875 ], "texts": [ "SUmmerMUsic", "BUSINESSINSPIRN DEATINS", "18Jan", "BUSINESSINSPIRATION Adds", "To construcon project is coand otheron project is colentcts ", "Call us :+098 123 4556 234", "Business Designs", "DJ Low-J & Yolk-X" ], "text_bbox": [ [ 196, 996, 1084, 1247 ], [ 711, 1604, 1198, 1729 ], [ 771, 294, 992, 493 ], [ 114, 1494, 542, 1591 ], [ 110, 1673, 629, 1730 ], [ 132, 104, 419, 169 ], [ 116, 1611, 553, 1648 ], [ 121, 1749, 532, 1786 ] ], "bbox_areas": [ 222888, 60875, 43979, 41516, 29583, 18655, 16169, 15207 ] }, { "task": "poster detection", "subtask": "", "name": "ec1de1972d4f40c69d0b8ea8639145d7.png", "path": "poster_ocr_1024/ec1de1972d4f40c69d0b8ea8639145d7.png", "path_original": "poster_ocr/ec1de1972d4f40c69d0b8ea8639145d7.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['Everlasting Grace Church FlyerBRBD | JOHN DOE | STEPHEN JACK | ALEX DIN | HAIDER ', 'Grace', 'SPECAIL GUESTDJ ALEXA | DJ JOHN', 'JUNUARY 25-12-2014-8PM BEST PARTY EVER', 'CHURCH', 'PRESENTANNAAYA-220', 'Everlasting']", "size": [ 1275, 1875 ], "texts": [ "Everlasting Grace Church FlyerBRBD | JOHN DOE | STEPHEN JACK | ALEX DIN | HAIDER ", "Grace", "SPECAIL GUESTDJ ALEXA | DJ JOHN", "JUNUARY 25-12-2014-8PM BEST PARTY EVER", "CHURCH", "PRESENTANNAAYA-220", "Everlasting" ], "text_bbox": [ [ 157, 1672, 1109, 1751 ], [ 356, 866, 908, 987 ], [ 471, 1103, 815, 1166 ], [ 238, 1766, 1035, 1792 ], [ 483, 1013, 800, 1071 ], [ 515, 705, 770, 766 ], [ 474, 808, 803, 845 ] ], "bbox_areas": [ 75208, 66792, 21672, 20722, 18386, 15555, 12173 ] }, { "task": "poster detection", "subtask": "", "name": "ec181163ef5443258cdee986a389485c.png", "path": "poster_ocr_1024/ec181163ef5443258cdee986a389485c.png", "path_original": "poster_ocr/ec181163ef5443258cdee986a389485c.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['GIRLs', 'genesis', 'dj NEON | Dj EDM | DJ MINIMAL', '9 PM . ENTRY 20$ . DRESS CODE : FUTURISTIC FANTASY', 'FRI15thJULY', 'WWW.YOURWEBSITE.COM', 'FLUO EVENTPRESENTS', 'f facebook.com/YOURNAME', 'T TWITTER.COM/YOURNAME']", "size": [ 1275, 1875 ], "texts": [ "GIRLs", "genesis", "dj NEON | Dj EDM | DJ MINIMAL", "9 PM . ENTRY 20$ . DRESS CODE : FUTURISTIC FANTASY", "FRI15thJULY", "WWW.YOURWEBSITE.COM", "FLUO EVENTPRESENTS", "f facebook.com/YOURNAME", "T TWITTER.COM/YOURNAME" ], "text_bbox": [ [ 230, 1174, 1018, 1375 ], [ 309, 1389, 940, 1473 ], [ 208, 1530, 1072, 1572 ], [ 200, 1592, 1075, 1618 ], [ 1057, 79, 1196, 230 ], [ 231, 1755, 1047, 1779 ], [ 74, 79, 301, 144 ], [ 219, 1621, 644, 1642 ], [ 660, 1621, 1056, 1642 ] ], "bbox_areas": [ 158388, 53004, 36288, 22750, 20989, 19584, 14755, 8925, 8316 ] }, { "task": "poster detection", "subtask": "", "name": "ec13c9f3af8041e3b0a4c6a2637a1661.png", "path": "poster_ocr_1024/ec13c9f3af8041e3b0a4c6a2637a1661.png", "path_original": "poster_ocr/ec13c9f3af8041e3b0a4c6a2637a1661.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['KEEPCALM', 'DESIGNFLYERS', '^', 'AND', 'www.flyerheroes.com']", "size": [ 1783, 2516 ], "texts": [ "KEEPCALM", "DESIGNFLYERS", "^", "AND", "www.flyerheroes.com" ], "text_bbox": [ [ 410, 735, 1361, 1298 ], [ 366, 1685, 1410, 2104 ], [ 618, 126, 1154, 592 ], [ 665, 1418, 1111, 1557 ], [ 561, 2300, 1221, 2342 ] ], "bbox_areas": [ 535413, 437436, 249776, 61994, 27720 ] }, { "task": "poster detection", "subtask": "", "name": "eb18621c674a4f9990c4c9c2ab7c54a6.png", "path": "poster_ocr_1024/eb18621c674a4f9990c4c9c2ab7c54a6.png", "path_original": "poster_ocr/eb18621c674a4f9990c4c9c2ab7c54a6.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['GOLDDRINKZ', 'dj SKULL | Dj MONEY | DJ STRIKE', '9 PM || ENTRY 20$ || DRESS CODE : ALL IN BLACK', '-OVNI EVENT PRESENTs-', 'NightClub', 'FRIDAY 15th JULY', 'f facebook.com/YOURNAME', 'T twitter.com/YOURNAME']", "size": [ 1275, 1875 ], "texts": [ "GOLDDRINKZ", "dj SKULL | Dj MONEY | DJ STRIKE", "9 PM || ENTRY 20$ || DRESS CODE : ALL IN BLACK", "-OVNI EVENT PRESENTs-", "NightClub", "FRIDAY 15th JULY", "f facebook.com/YOURNAME", "T twitter.com/YOURNAME" ], "text_bbox": [ [ 206, 1171, 1075, 1481 ], [ 241, 1530, 1033, 1582 ], [ 239, 1601, 1038, 1636 ], [ 277, 1100, 991, 1126 ], [ 525, 1652, 762, 1707 ], [ 491, 1044, 777, 1083 ], [ 254, 1652, 522, 1681 ], [ 772, 1652, 1019, 1681 ] ], "bbox_areas": [ 269390, 41184, 27965, 18564, 13035, 11154, 7772, 7163 ] }, { "task": "poster detection", "subtask": "", "name": "eb46f868a9b9419dacf3ba9772c776ce.png", "path": "poster_ocr_1024/eb46f868a9b9419dacf3ba9772c776ce.png", "path_original": "poster_ocr/eb46f868a9b9419dacf3ba9772c776ce.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n[' SHISHAHOUSE', 'dj BLACK | Dj HOOKAH | DJ SMOKE', '9 PM . ENTRY 20$ . DRESS CODE . ALL IN BLACK', '-OVNI EVENT PRESENTs-', 'NightClub', 'FRIDAY 15th JULY', 'f facebook.com/YOURNAME', 'T twitter.com/YOURNAME']", "size": [ 1275, 1875 ], "texts": [ " SHISHAHOUSE", "dj BLACK | Dj HOOKAH | DJ SMOKE", "9 PM . ENTRY 20$ . DRESS CODE . ALL IN BLACK", "-OVNI EVENT PRESENTs-", "NightClub", "FRIDAY 15th JULY", "f facebook.com/YOURNAME", "T twitter.com/YOURNAME" ], "text_bbox": [ [ 183, 1206, 1089, 1485 ], [ 160, 1529, 1120, 1587 ], [ 195, 1606, 1082, 1639 ], [ 211, 1129, 1064, 1159 ], [ 510, 1663, 778, 1724 ], [ 466, 1062, 808, 1107 ], [ 203, 1663, 506, 1695 ], [ 789, 1663, 1069, 1695 ] ], "bbox_areas": [ 252774, 55680, 29271, 25590, 16348, 15390, 9696, 8960 ] }, { "task": "poster detection", "subtask": "", "name": "ea2c2812851d4709a90a9fd0ff866ccf.png", "path": "poster_ocr_1024/ea2c2812851d4709a90a9fd0ff866ccf.png", "path_original": "poster_ocr/ea2c2812851d4709a90a9fd0ff866ccf.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['DARK SHADOW ', 'DIGITAL MOTION PICTURES PRESENTS THE BEST DESIGN ARTINSPIRATIONS and VISIONS reserved 2015 COPYRIGHTS', 'KYLE HURT / JULIA STAN / BRUCE WYLYS', '-OVNI EVENT PRESENTs-', 'WEBSITE COMING SOON', 'F WWW.YOURFACEBOOK.COM', '18.9.15', 'T www.YOURTWITTER.com', 'YOURPLACE']", "size": [ 1275, 1875 ], "texts": [ "DARK SHADOW ", "DIGITAL MOTION PICTURES PRESENTS THE BEST DESIGN ARTINSPIRATIONS and VISIONS reserved 2015 COPYRIGHTS", "KYLE HURT / JULIA STAN / BRUCE WYLYS", "-OVNI EVENT PRESENTs-", "WEBSITE COMING SOON", "F WWW.YOURFACEBOOK.COM", "18.9.15", "T www.YOURTWITTER.com", "YOURPLACE" ], "text_bbox": [ [ 270, 1060, 1001, 1277 ], [ 273, 1526, 1011, 1620 ], [ 275, 1383, 1005, 1421 ], [ 288, 986, 981, 1014 ], [ 303, 1768, 945, 1789 ], [ 737, 1646, 1026, 1674 ], [ 540, 1707, 746, 1746 ], [ 256, 1646, 533, 1674 ], [ 544, 1645, 728, 1682 ] ], "bbox_areas": [ 158627, 69372, 27740, 19404, 13482, 8092, 8034, 7756, 6808 ] }, { "task": "poster detection", "subtask": "", "name": "ea381f6f36254afcab5d79a0387fd835.png", "path": "poster_ocr_1024/ea381f6f36254afcab5d79a0387fd835.png", "path_original": "poster_ocr/ea381f6f36254afcab5d79a0387fd835.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['Workshop', 'Gallery', 'Office', 'Hello World', 'swipe for more']", "size": [ 1000, 1000 ], "texts": [ "Workshop", "Gallery", "Office", "Hello World", "swipe for more" ], "text_bbox": [ [ 63, 124, 380, 219 ], [ 541, 348, 779, 439 ], [ 172, 578, 374, 661 ], [ 353, 49, 648, 85 ], [ 387, 938, 614, 953 ] ], "bbox_areas": [ 30115, 21658, 16766, 10620, 3405 ] }, { "task": "poster detection", "subtask": "", "name": "e961714977bd42b8bef97382e2daca11.png", "path": "poster_ocr_1024/e961714977bd42b8bef97382e2daca11.png", "path_original": "poster_ocr/e961714977bd42b8bef97382e2daca11.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['CHRISTMAS', 'PARTY', 'DECEMBER', '15$ TICKETS - FREE ENTRY FOR KIDS', 'REMEMBER TO BRING TOYS FOR THE TOY DRIVE', '24TH', 'FAMILY FUN - MANY PRESENTS - CONTESTS', 'EVENT ORGANISED BY ZEPPELIN PUB & RADIO ONE', 'TUESDAY', 'ZEPPELIN PRESENTS', 'WWW.WEBSITE.COM', 'DOORS OPEN AT 6PM', '0931 MILLS STREET, CALIFORNIA - NEAR THE PLAZA']", "size": [ 2700, 3450 ], "texts": [ "CHRISTMAS", "PARTY", "DECEMBER", "15$ TICKETS - FREE ENTRY FOR KIDS", "REMEMBER TO BRING TOYS FOR THE TOY DRIVE", "24TH", "FAMILY FUN - MANY PRESENTS - CONTESTS", "EVENT ORGANISED BY ZEPPELIN PUB & RADIO ONE", "TUESDAY", "ZEPPELIN PRESENTS", "WWW.WEBSITE.COM", "DOORS OPEN AT 6PM", "0931 MILLS STREET, CALIFORNIA - NEAR THE PLAZA" ], "text_bbox": [ [ 317, 544, 2375, 994 ], [ 793, 1022, 1898, 1458 ], [ 805, 1702, 1888, 1959 ], [ 316, 2603, 2375, 2701 ], [ 490, 2782, 2208, 2870 ], [ 864, 1996, 1544, 2190 ], [ 315, 2523, 2374, 2579 ], [ 504, 1559, 2185, 1599 ], [ 1103, 2228, 1773, 2321 ], [ 852, 420, 1848, 467 ], [ 956, 3068, 1740, 3118 ], [ 969, 2380, 1726, 2427 ], [ 788, 2975, 1909, 3004 ] ], "bbox_areas": [ 926100, 481780, 278331, 201782, 151184, 131920, 115304, 67240, 62310, 46812, 39200, 35579, 32509 ] }, { "task": "poster detection", "subtask": "", "name": "e94821fc076042e1a280c5ad22409240.png", "path": "poster_ocr_1024/e94821fc076042e1a280c5ad22409240.png", "path_original": "poster_ocr/e94821fc076042e1a280c5ad22409240.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['INDIEROCK', 'MAROON, ANTZ& SAVAGE.', 'TONIGHT', 'MAY 5TH 2015', 'FOR MORE INFO VISIT: WWW.WEBSITE.COM', '35$ TICKETS, 21+ TO ENTER', ' SPECIAL GUESTS', 'FEATURING', 'OR CALL US AT: 5522 225 255', 'ZEPPELIN PRESENTS', 'FREE TABLE & PARKING', 'FACEBOOK/INDIEROCK', '0923, MAIN STREET, YOUR CITY CA. NEAR THE PLAZA']", "size": [ 2700, 3450 ], "texts": [ "INDIEROCK", "MAROON, ANTZ& SAVAGE.", "TONIGHT", "MAY 5TH 2015", "FOR MORE INFO VISIT: WWW.WEBSITE.COM", "35$ TICKETS, 21+ TO ENTER", " SPECIAL GUESTS", "FEATURING", "OR CALL US AT: 5522 225 255", "ZEPPELIN PRESENTS", "FREE TABLE & PARKING", "FACEBOOK/INDIEROCK", "0923, MAIN STREET, YOUR CITY CA. NEAR THE PLAZA" ], "text_bbox": [ [ 222, 322, 2477, 878 ], [ 219, 1908, 2040, 2263 ], [ 219, 897, 1883, 1156 ], [ 289, 1259, 1454, 1510 ], [ 221, 2635, 1995, 2693 ], [ 223, 2429, 1307, 2510 ], [ 221, 1782, 1164, 1862 ], [ 425, 1681, 1164, 1776 ], [ 222, 2704, 1367, 2762 ], [ 327, 196, 1299, 262 ], [ 223, 2511, 1174, 2571 ], [ 268, 2961, 920, 3043 ], [ 306, 2824, 1616, 2863 ] ], "bbox_areas": [ 1253780, 646455, 430976, 292415, 102892, 87804, 75440, 70205, 66410, 64152, 57060, 53464, 51090 ] }, { "task": "poster detection", "subtask": "", "name": "e924ed36d6e94265940389c9db8c10aa.png", "path": "poster_ocr_1024/e924ed36d6e94265940389c9db8c10aa.png", "path_original": "poster_ocr/e924ed36d6e94265940389c9db8c10aa.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['BOB MARLEY | KURT COBAIN | JOHN LENNON | ELVIS PRESLEY', 'AMY WINEHOUSE | freddie mercury | JIM MORRISON', '- LOREM IPSUM DOLOR SIT AMET -', ' Lorem ipsum dolor sit amet, consectetuer adipiscing elit, sed diam nonummy nibh euismod tincidunt ut laoreet dolore magna aliquam erat volutpat. Ut wisi enim ad minim veniam, quis nostrud exerci tation ullamcorper suscipit lobortis nisl ut aliquip ex ea commodo consequat. Duis autem vel eum iriure dolor in hendrerit.', 'FRIDAY10th JAN', '15$ ENTRY', 'YOUR CLUB PRESENT', 'www.yourclub.com']", "size": [ 2570, 3598 ], "texts": [ "BOB MARLEY | KURT COBAIN | JOHN LENNON | ELVIS PRESLEY", "AMY WINEHOUSE | freddie mercury | JIM MORRISON", "- LOREM IPSUM DOLOR SIT AMET -", " Lorem ipsum dolor sit amet, consectetuer adipiscing elit, sed diam nonummy nibh euismod tincidunt ut laoreet dolore magna aliquam erat volutpat. Ut wisi enim ad minim veniam, quis nostrud exerci tation ullamcorper suscipit lobortis nisl ut aliquip ex ea commodo consequat. Duis autem vel eum iriure dolor in hendrerit.", "FRIDAY10th JAN", "15$ ENTRY", "YOUR CLUB PRESENT", "www.yourclub.com" ], "text_bbox": [ [ 337, 2776, 2232, 2857 ], [ 460, 2886, 2110, 2967 ], [ 714, 905, 1824, 1019 ], [ 396, 3027, 2227, 3093 ], [ 1673, 1797, 2276, 1940 ], [ 458, 1792, 869, 1940 ], [ 1016, 262, 1554, 328 ], [ 1060, 3148, 1508, 3201 ] ], "bbox_areas": [ 153495, 133650, 126540, 120846, 86229, 60828, 35508, 23744 ] }, { "task": "poster detection", "subtask": "", "name": "e8b1b8f234ca421587f10f4ccd279cf4.png", "path": "poster_ocr_1024/e8b1b8f234ca421587f10f4ccd279cf4.png", "path_original": "poster_ocr/e8b1b8f234ca421587f10f4ccd279cf4.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['SUMMER', 'BEACHPARADISE', '27', 'JULY']", "size": [ 3543, 4961 ], "texts": [ "SUMMER", "BEACHPARADISE", "27", "JULY" ], "text_bbox": [ [ 1151, 1313, 2349, 3431 ], [ 1507, 703, 1970, 878 ], [ 3014, 3210, 3146, 3301 ], [ 3006, 3328, 3151, 3376 ] ], "bbox_areas": [ 2537364, 81025, 12012, 6960 ] }, { "task": "poster detection", "subtask": "", "name": "e82f553c48e54513a6a7987d51ae0336.png", "path": "poster_ocr_1024/e82f553c48e54513a6a7987d51ae0336.png", "path_original": "poster_ocr/e82f553c48e54513a6a7987d51ae0336.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['Mindful', '7', 'days plan', 'Create Your Own Detox Plan', '*Official Launch: AUG 15TH', 'YES, PLEASE!', 'DETOX']", "size": [ 1080, 1920 ], "texts": [ "Mindful", "7", "days plan", "Create Your Own Detox Plan", "*Official Launch: AUG 15TH", "YES, PLEASE!", "DETOX" ], "text_bbox": [ [ 290, 775, 783, 1091 ], [ 465, 1260, 619, 1471 ], [ 190, 1368, 891, 1402 ], [ 192, 1159, 886, 1186 ], [ 330, 1645, 749, 1669 ], [ 378, 1552, 700, 1582 ], [ 451, 1048, 635, 1082 ] ], "bbox_areas": [ 155788, 32494, 23834, 18738, 10056, 9660, 6256 ] }, { "task": "poster detection", "subtask": "", "name": "e827f7403b584086a9f3be51969d6a90.png", "path": "poster_ocr_1024/e827f7403b584086a9f3be51969d6a90.png", "path_original": "poster_ocr/e827f7403b584086a9f3be51969d6a90.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['SUMMER', 'SATURDAY', '- BEACH PARTY -', '28TH AUGUST', 'FACEBOOK/YOUREVENT TWITTER/YOUREVENT YOUTUBE/YOUREVENT', 'WWW.WEBSITE.COM', 'FREE ENTRY | DOORS OPEN AT 9PM', '- YOURCLUB PRESENTS -', '- FEATURING DJ SLICK & DOZER -', '8214 MAIN STREET YOUR CITY CA. NEAR THE PLAZA']", "size": [ 2700, 3450 ], "texts": [ "SUMMER", "SATURDAY", "- BEACH PARTY -", "28TH AUGUST", "FACEBOOK/YOUREVENT TWITTER/YOUREVENT YOUTUBE/YOUREVENT", "WWW.WEBSITE.COM", "FREE ENTRY | DOORS OPEN AT 9PM", "- YOURCLUB PRESENTS -", "- FEATURING DJ SLICK & DOZER -", "8214 MAIN STREET YOUR CITY CA. NEAR THE PLAZA" ], "text_bbox": [ [ 797, 409, 1903, 912 ], [ 1066, 1154, 1631, 1383 ], [ 794, 929, 1905, 1038 ], [ 895, 1390, 1808, 1497 ], [ 342, 3289, 2416, 3322 ], [ 1075, 1775, 1620, 1880 ], [ 846, 1661, 1851, 1717 ], [ 979, 268, 1718, 332 ], [ 915, 1605, 1782, 1658 ], [ 942, 1907, 1749, 1932 ] ], "bbox_areas": [ 556318, 129385, 121099, 97691, 68442, 57225, 56280, 47296, 45951, 20175 ] }, { "task": "poster detection", "subtask": "", "name": "e813d96fb00e42239e7a02a263441dd0.png", "path": "poster_ocr_1024/e813d96fb00e42239e7a02a263441dd0.png", "path_original": "poster_ocr/e813d96fb00e42239e7a02a263441dd0.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['grid', 'Peace and Love', 'www.domainaddress.com', 'imagination', 'availability', 'Inspiration', 'Innovation', 'Expensive', 'Vintage']", "size": [ 1200, 1200 ], "texts": [ "grid", "Peace and Love", "www.domainaddress.com", "imagination", "availability", "Inspiration", "Innovation", "Expensive", "Vintage" ], "text_bbox": [ [ 91, 467, 161, 734 ], [ 478, 1076, 722, 1112 ], [ 454, 71, 748, 89 ], [ 1008, 354, 1152, 368 ], [ 1009, 834, 1151, 848 ], [ 532, 354, 668, 368 ], [ 533, 834, 667, 848 ], [ 303, 594, 418, 608 ], [ 794, 594, 886, 608 ] ], "bbox_areas": [ 18690, 8784, 5292, 2016, 1988, 1904, 1876, 1610, 1288 ] }, { "task": "poster detection", "subtask": "", "name": "e7c946c7c46e4763a137192daed0802a.png", "path": "poster_ocr_1024/e7c946c7c46e4763a137192daed0802a.png", "path_original": "poster_ocr/e7c946c7c46e4763a137192daed0802a.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['CIGAR & WHISKEY', 'dj SKULL | Dj MONEY | DJ STRIKE', '9 PM || ENTRY 20$ || DRESS CODE : ALL IN BLACK', '-OVNI EVENT PRESENTs-', 'NightClub', 'FRIDAY 15th JULY', 'f facebook.com/YOURNAME', 'T twitter.com/YOURNAME']", "size": [ 1275, 1875 ], "texts": [ "CIGAR & WHISKEY", "dj SKULL | Dj MONEY | DJ STRIKE", "9 PM || ENTRY 20$ || DRESS CODE : ALL IN BLACK", "-OVNI EVENT PRESENTs-", "NightClub", "FRIDAY 15th JULY", "f facebook.com/YOURNAME", "T twitter.com/YOURNAME" ], "text_bbox": [ [ 186, 1193, 1104, 1457 ], [ 241, 1530, 1033, 1582 ], [ 239, 1601, 1038, 1636 ], [ 277, 1100, 991, 1126 ], [ 525, 1652, 762, 1707 ], [ 491, 1044, 777, 1083 ], [ 254, 1652, 522, 1681 ], [ 772, 1652, 1019, 1681 ] ], "bbox_areas": [ 242352, 41184, 27965, 18564, 13035, 11154, 7772, 7163 ] }, { "task": "poster detection", "subtask": "", "name": "e71369c6737f4aa28f9f6e591c4e5923.png", "path": "poster_ocr_1024/e71369c6737f4aa28f9f6e591c4e5923.png", "path_original": "poster_ocr/e71369c6737f4aa28f9f6e591c4e5923.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['GOTHIC ACTION', 'HERO', 'YOUR LINE GOSE TO HERE', '2016 Revolution']", "size": [ 3000, 4000 ], "texts": [ "GOTHIC ACTION", "HERO", "YOUR LINE GOSE TO HERE", "2016 Revolution" ], "text_bbox": [ [ 241, 3213, 1677, 3349 ], [ 913, 2981, 1676, 3198 ], [ 236, 3388, 1681, 3473 ], [ 1039, 3500, 1688, 3570 ] ], "bbox_areas": [ 195296, 165571, 122825, 45430 ] }, { "task": "poster detection", "subtask": "", "name": "e6a94047e28048cbad09ffa0fbae68cf.png", "path": "poster_ocr_1024/e6a94047e28048cbad09ffa0fbae68cf.png", "path_original": "poster_ocr/e6a94047e28048cbad09ffa0fbae68cf.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['GARDEN', ' ADMISSION | 123 456 789 | 1 FREE DRINKSTREET NAME 123 | CITY', '23 MAY ', 'party', 'rd']", "size": [ 3333, 3333 ], "texts": [ "GARDEN", " ADMISSION | 123 456 789 | 1 FREE DRINKSTREET NAME 123 | CITY", "23 MAY ", "party", "rd" ], "text_bbox": [ [ 866, 1139, 2472, 1466 ], [ 772, 2241, 2579, 2398 ], [ 1156, 1759, 2120, 1916 ], [ 1319, 1536, 2018, 1635 ], [ 1412, 1734, 1530, 1813 ] ], "bbox_areas": [ 525162, 283699, 151348, 69201, 9322 ] }, { "task": "poster detection", "subtask": "", "name": "e6830b44382045ac99bb87893ea645a8.png", "path": "poster_ocr_1024/e6830b44382045ac99bb87893ea645a8.png", "path_original": "poster_ocr/e6830b44382045ac99bb87893ea645a8.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['ELECTRO', 'NIGHTS', 'MORE TEXT HERE OR DETAILS IF NECESSSARY. SPONOSORS, DETAILS ETC. OR DELETE THE TEXT AND USE YOUR LOGO.DELETE THE TEXT AND USE YOUR LOGO HERE. MRE TEXT OR DETAILS IF NECESSARY, SPONSORS, INFO.', '20$ ENTRY | FREE FOR GIRLS | 5$ DRINKS | VALET PARKING ', 'HOSTED BY MC DAVEY // FOR MORE INFO: WWW.WEBSITE.COM OR CALL US: 555 666 333', '@ CLUB ZEPPELIN', '18943 MAIN STREET, CITY, CA. NEAR THE PLAZA', 'SUMMERFEST', 'AUGUST 21ST', '2014EDITION', 'DOORS OPEN AT 22PM', 'SATURDAY', 'DJ SHADE', 'ZEPPELIN PRESENTS', 'DJ MARS', 'ELECTRONIC MUSIC FETIVAL', 'DJ SLICK', 'DJ PAUL', 'FACEBOOK', 'YOUTUBE', 'TWITTER', 'VIMEO']", "size": [ 2700, 3450 ], "texts": [ "ELECTRO", "NIGHTS", "MORE TEXT HERE OR DETAILS IF NECESSSARY. SPONOSORS, DETAILS ETC. OR DELETE THE TEXT AND USE YOUR LOGO.DELETE THE TEXT AND USE YOUR LOGO HERE. MRE TEXT OR DETAILS IF NECESSARY, SPONSORS, INFO.", "20$ ENTRY | FREE FOR GIRLS | 5$ DRINKS | VALET PARKING ", "HOSTED BY MC DAVEY // FOR MORE INFO: WWW.WEBSITE.COM OR CALL US: 555 666 333", "@ CLUB ZEPPELIN", "18943 MAIN STREET, CITY, CA. NEAR THE PLAZA", "SUMMERFEST", "AUGUST 21ST", "2014EDITION", "DOORS OPEN AT 22PM", "SATURDAY", "DJ SHADE", "ZEPPELIN PRESENTS", "DJ MARS", "ELECTRONIC MUSIC FETIVAL", "DJ SLICK", "DJ PAUL", "FACEBOOK", "YOUTUBE", "TWITTER", "VIMEO" ], "text_bbox": [ [ 751, 1387, 1956, 1634 ], [ 863, 1658, 1848, 1878 ], [ 357, 3059, 2345, 3136 ], [ 356, 2810, 2340, 2874 ], [ 357, 438, 2341, 479 ], [ 917, 2572, 1774, 2656 ], [ 665, 2891, 2038, 2943 ], [ 962, 542, 1729, 630 ], [ 1041, 2075, 1654, 2144 ], [ 1202, 1079, 1497, 1209 ], [ 1006, 2659, 1687, 2707 ], [ 1119, 2006, 1573, 2071 ], [ 879, 295, 1275, 359 ], [ 1067, 158, 1630, 201 ], [ 361, 296, 718, 360 ], [ 1033, 645, 1656, 681 ], [ 1467, 295, 1811, 359 ], [ 1996, 299, 2342, 361 ], [ 428, 3265, 701, 3303 ], [ 2103, 3263, 2345, 3301 ], [ 1048, 3267, 1272, 3303 ], [ 1615, 3265, 1776, 3303 ] ], "bbox_areas": [ 297635, 216700, 153076, 126976, 81344, 71988, 71396, 67496, 42297, 38350, 32688, 29510, 25344, 24209, 22848, 22428, 22016, 21452, 10374, 9196, 8064, 6118 ] }, { "task": "poster detection", "subtask": "", "name": "e6715ec83be3469990c0dca80f823b77.png", "path": "poster_ocr_1024/e6715ec83be3469990c0dca80f823b77.png", "path_original": "poster_ocr/e6715ec83be3469990c0dca80f823b77.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['Silver', 'Gold', '&', 'Your Brand', 'SATURDAY 25th OCTOBER 2014', 'Club designers 123 St. YOUR CITY', 'ENTRY 10$ free admission for ladies until 10', 'YOUR ENTERTAINMENT PRESENTS', 'NightClub', 'facebook.com/YOURNAME', 'twitter.com/YOURNAME', 'www.clubname.com']", "size": [ 1275, 1875 ], "texts": [ "Silver", "Gold", "&", "Your Brand", "SATURDAY 25th OCTOBER 2014", "Club designers 123 St. YOUR CITY", "ENTRY 10$ free admission for ladies until 10", "YOUR ENTERTAINMENT PRESENTS", "NightClub", "facebook.com/YOURNAME", "twitter.com/YOURNAME", "www.clubname.com" ], "text_bbox": [ [ 386, 862, 1152, 1132 ], [ 378, 590, 1024, 860 ], [ 265, 665, 450, 893 ], [ 233, 884, 326, 1274 ], [ 458, 460, 1018, 517 ], [ 454, 1222, 1073, 1264 ], [ 455, 1270, 1071, 1292 ], [ 471, 424, 1006, 448 ], [ 667, 1312, 861, 1369 ], [ 448, 1315, 650, 1336 ], [ 885, 1315, 1072, 1336 ], [ 680, 1367, 844, 1385 ] ], "bbox_areas": [ 206820, 174420, 42180, 36270, 31920, 25998, 13552, 12840, 11058, 4242, 3927, 2952 ] }, { "task": "poster detection", "subtask": "", "name": "e4c574959668488886a81d3d76ab5f0d.png", "path": "poster_ocr_1024/e4c574959668488886a81d3d76ab5f0d.png", "path_original": "poster_ocr/e4c574959668488886a81d3d76ab5f0d.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['FUTUR', 'Boombox', 'dj LOVER | Dj CLASS | DJ GRAPHICS', 'Club designers 123 St. YOUR CITY', 'ADMISSION 10$ | FREE FOR LADIES | START 7:30 PM', 'YOUR ENTERTAINMENT PRESENTS', '25.10.2015', 'F facebook.com/YOURNAME', 'T twitter.com/YOURNAME', 'W wWw.website.com']", "size": [ 1275, 1875 ], "texts": [ "FUTUR", "Boombox", "dj LOVER | Dj CLASS | DJ GRAPHICS", "Club designers 123 St. YOUR CITY", "ADMISSION 10$ | FREE FOR LADIES | START 7:30 PM", "YOUR ENTERTAINMENT PRESENTS", "25.10.2015", "F facebook.com/YOURNAME", "T twitter.com/YOURNAME", "W wWw.website.com" ], "text_bbox": [ [ 255, 179, 1025, 511 ], [ 379, 424, 882, 665 ], [ 240, 1485, 1011, 1549 ], [ 240, 1553, 1011, 1601 ], [ 265, 1619, 988, 1648 ], [ 264, 138, 991, 161 ], [ 498, 1414, 748, 1462 ], [ 261, 1649, 501, 1675 ], [ 541, 1649, 762, 1675 ], [ 800, 1649, 988, 1674 ] ], "bbox_areas": [ 255640, 121223, 49344, 37008, 20967, 16721, 12000, 6240, 5746, 4700 ] }, { "task": "poster detection", "subtask": "", "name": "e4bb3c2aaa8f44f9ab0443d9b5c8b933.png", "path": "poster_ocr_1024/e4bb3c2aaa8f44f9ab0443d9b5c8b933.png", "path_original": "poster_ocr/e4bb3c2aaa8f44f9ab0443d9b5c8b933.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['GOLDparty', 'dj GOLD | Dj STYLE | DJ LUXURY', 'SATURDAY 25th OCTOBER 2015', '@GoldClub', 'Club designers 123 St. YOUR CITY', 'ENTRY 10$ free admission for ladies until 10', 'YOUR ENTERTAINMENT PRESENTS', 'F facebook.com/YOURNAME', 'T twitter.com/YOURNAME', 'W YOURWEBSITE.COM']", "size": [ 1275, 1875 ], "texts": [ "GOLDparty", "dj GOLD | Dj STYLE | DJ LUXURY", "SATURDAY 25th OCTOBER 2015", "@GoldClub", "Club designers 123 St. YOUR CITY", "ENTRY 10$ free admission for ladies until 10", "YOUR ENTERTAINMENT PRESENTS", "F facebook.com/YOURNAME", "T twitter.com/YOURNAME", "W YOURWEBSITE.COM" ], "text_bbox": [ [ 316, 387, 957, 1252 ], [ 347, 1290, 936, 1356 ], [ 366, 288, 902, 343 ], [ 500, 1582, 758, 1683 ], [ 347, 1358, 933, 1400 ], [ 333, 1405, 947, 1428 ], [ 334, 210, 936, 229 ], [ 347, 1451, 558, 1474 ], [ 734, 1451, 929, 1474 ], [ 566, 1451, 724, 1473 ] ], "bbox_areas": [ 554465, 38874, 29480, 26058, 24612, 14122, 11438, 4853, 4485, 3476 ] }, { "task": "poster detection", "subtask": "", "name": "e441b28988f74425b936215cb4568396.png", "path": "poster_ocr_1024/e441b28988f74425b936215cb4568396.png", "path_original": "poster_ocr/e441b28988f74425b936215cb4568396.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['deep', 'sounds', 'SEBASTIAN', 'GIGAMESH & EXAMPLE', '///////// 29.08.13', 'A Nightclub Special Event', '10 $ ENTRY AND FREE DRINKS UNTIL MIDNIGHT', 'DOORS OPEN AT 11 PM', 'INDIEGROUND PRESENTS', 'FEATURING', 'INDIEGROUND.IT']", "size": [ 2539, 3567 ], "texts": [ "deep", "sounds", "SEBASTIAN", "GIGAMESH & EXAMPLE", "///////// 29.08.13", "A Nightclub Special Event", "10 $ ENTRY AND FREE DRINKS UNTIL MIDNIGHT", "DOORS OPEN AT 11 PM", "INDIEGROUND PRESENTS", "FEATURING", "INDIEGROUND.IT" ], "text_bbox": [ [ 545, 922, 2108, 1352 ], [ 335, 1348, 1812, 1601 ], [ 898, 2068, 2153, 2281 ], [ 711, 2297, 1928, 2399 ], [ 613, 1636, 1371, 1790 ], [ 901, 1833, 1988, 1936 ], [ 866, 2614, 1780, 2661 ], [ 1028, 2548, 1702, 2606 ], [ 1194, 861, 1680, 902 ], [ 1490, 1973, 1959, 2015 ], [ 1044, 2691, 1392, 2721 ] ], "bbox_areas": [ 672090, 373681, 267315, 124134, 116732, 111961, 42958, 39092, 19926, 19698, 10440 ] }, { "task": "poster detection", "subtask": "", "name": "e43c81198dbf4378ace6aed3f49a373a.png", "path": "poster_ocr_1024/e43c81198dbf4378ace6aed3f49a373a.png", "path_original": "poster_ocr/e43c81198dbf4378ace6aed3f49a373a.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['WINTER BASH', 'FEAT. DJ SLICK', 'WINTER SHOWDOWN', '5 STAGES 8 BARS FREE PARKING 21+ TO ENTER', 'DECEMBER', '50$ TICKETS FOR THE ENTIRE FESTIVAL 20$/DAY', '26-29', 'FOR RESERVATION CALL 5544 445 445', 'ZEPPELIN', '& RADIO 54', 'MC DAVE', 'WWW.WEBSITE.COM', '& MANY MORE GUESTS', 'FESTIVAL', 'RONSON FIVE DJ CLUE', 'CAMILLE FIVE AARON 6', 'PRISMATIC DUB SENSES', '0921 STATE DRIVE NEW YORK CITY, NY, NEAR THE PLAZA', 'FACEBOOK/WINTERBASH TWITTER@WINTERBASH', 'STARTS', 'DOORS OPEN AT 2PMEVERY DAY', 'BROUGHTTO YOU BY']", "size": [ 2700, 3450 ], "texts": [ "WINTER BASH", "FEAT. DJ SLICK", "WINTER SHOWDOWN", "5 STAGES 8 BARS FREE PARKING 21+ TO ENTER", "DECEMBER", "50$ TICKETS FOR THE ENTIRE FESTIVAL 20$/DAY", "26-29", "FOR RESERVATION CALL 5544 445 445", "ZEPPELIN", "& RADIO 54", "MC DAVE", "WWW.WEBSITE.COM", "& MANY MORE GUESTS", "FESTIVAL", "RONSON FIVE DJ CLUE", "CAMILLE FIVE AARON 6", "PRISMATIC DUB SENSES", "0921 STATE DRIVE NEW YORK CITY, NY, NEAR THE PLAZA", "FACEBOOK/WINTERBASH TWITTER@WINTERBASH", "STARTS", "DOORS OPEN AT 2PMEVERY DAY", "BROUGHTTO YOU BY" ], "text_bbox": [ [ 130, 1934, 2576, 2437 ], [ 170, 2640, 1311, 2872 ], [ 122, 2475, 1825, 2588 ], [ 303, 3019, 2072, 3114 ], [ 248, 488, 999, 690 ], [ 300, 2947, 2073, 3011 ], [ 209, 297, 827, 454 ], [ 1076, 3187, 2575, 3244 ], [ 2061, 1288, 2497, 1426 ], [ 1837, 1436, 2499, 1523 ], [ 2136, 2971, 2555, 3099 ], [ 146, 3164, 1006, 3223 ], [ 1401, 2841, 2290, 2898 ], [ 1913, 2492, 2544, 2570 ], [ 1401, 2649, 2290, 2704 ], [ 1401, 2777, 2291, 2830 ], [ 1401, 2716, 2291, 2768 ], [ 122, 3269, 1345, 3304 ], [ 122, 3303, 1082, 3338 ], [ 205, 213, 588, 286 ], [ 206, 731, 621, 796 ], [ 2263, 1210, 2498, 1273 ] ], "bbox_areas": [ 1230338, 264712, 192439, 168055, 151702, 113472, 97026, 85443, 60168, 57594, 53632, 50740, 50673, 49218, 48895, 47170, 46280, 42805, 33600, 27959, 26975, 14805 ] }, { "task": "poster detection", "subtask": "", "name": "e37757fb2eba4ba1b6db54f84814f046.png", "path": "poster_ocr_1024/e37757fb2eba4ba1b6db54f84814f046.png", "path_original": "poster_ocr/e37757fb2eba4ba1b6db54f84814f046.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['LOVE', 'MEMORIES', 'Retro Card', 'VINTAGE STYLE FRAME']", "size": [ 3334, 3333 ], "texts": [ "LOVE", "MEMORIES", "Retro Card", "VINTAGE STYLE FRAME" ], "text_bbox": [ [ 1232, 1881, 2102, 2129 ], [ 1229, 2290, 2105, 2409 ], [ 1406, 1732, 1932, 1846 ], [ 1232, 2191, 2097, 2228 ] ], "bbox_areas": [ 215760, 104244, 59964, 32005 ] }, { "task": "poster detection", "subtask": "", "name": "e36e5db9e3f743ad8778d0a4cf829b4d.png", "path": "poster_ocr_1024/e36e5db9e3f743ad8778d0a4cf829b4d.png", "path_original": "poster_ocr/e36e5db9e3f743ad8778d0a4cf829b4d.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['expressyourself.', 'Lorem ipsum dolor sit amet, consectetur adipisc ing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua. Quis ipsum sus pendisse. ultrices gravi da. Risus commodo viverra maecenas. accumsan lacus vel facilisis. ', '20%', 'loremipsumdolor', 'Lorem Ipsum dolor --- sit amet consectetur', 'sitamet', 'off']", "size": [ 2480, 3508 ], "texts": [ "expressyourself.", "Lorem ipsum dolor sit amet, consectetur adipisc ing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua. Quis ipsum sus pendisse. ultrices gravi da. Risus commodo viverra maecenas. accumsan lacus vel facilisis. ", "20%", "loremipsumdolor", "Lorem Ipsum dolor --- sit amet consectetur", "sitamet", "off" ], "text_bbox": [ [ 163, 2541, 1479, 2961 ], [ 179, 3024, 1478, 3176 ], [ 1717, 2800, 2212, 3004 ], [ 360, 364, 691, 630 ], [ 180, 3314, 1466, 3357 ], [ 654, 754, 945, 926 ], [ 1835, 3027, 2072, 3117 ] ], "bbox_areas": [ 552720, 197448, 100980, 88046, 55298, 50052, 21330 ] }, { "task": "poster detection", "subtask": "", "name": "e2ca9fa78dae47a696c0452958f23ce5.png", "path": "poster_ocr_1024/e2ca9fa78dae47a696c0452958f23ce5.png", "path_original": "poster_ocr/e2ca9fa78dae47a696c0452958f23ce5.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['ILLUSTRATIONWorkshop', 'MELANIEDAVEID.COM', 'with Melania Daveid']", "size": [ 1080, 1080 ], "texts": [ "ILLUSTRATIONWorkshop", "MELANIEDAVEID.COM", "with Melania Daveid" ], "text_bbox": [ [ 231, 96, 850, 275 ], [ 310, 1020, 764, 1044 ], [ 383, 304, 699, 323 ] ], "bbox_areas": [ 110801, 10896, 6004 ] }, { "task": "poster detection", "subtask": "", "name": "e1e94e0b1794453ab2c6da5518270b67.png", "path": "poster_ocr_1024/e1e94e0b1794453ab2c6da5518270b67.png", "path_original": "poster_ocr/e1e94e0b1794453ab2c6da5518270b67.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['Stock mainstream tailored buttons allure hanger breathable old-fashioned outlet young garment look unique pattern.', 'Expensive adjustment sari look condition expirement model swag beautiful.', 'Peace and Love', 'www.domainaddress.com']", "size": [ 1200, 1200 ], "texts": [ "Stock mainstream tailored buttons allure hanger breathable old-fashioned outlet young garment look unique pattern.", "Expensive adjustment sari look condition expirement model swag beautiful.", "Peace and Love", "www.domainaddress.com" ], "text_bbox": [ [ 89, 164, 481, 339 ], [ 831, 743, 1094, 989 ], [ 89, 1076, 334, 1112 ], [ 89, 71, 383, 89 ] ], "bbox_areas": [ 68600, 64698, 8820, 5292 ] }, { "task": "poster detection", "subtask": "", "name": "e12ac268034545ba993c226252692cd6.png", "path": "poster_ocr_1024/e12ac268034545ba993c226252692cd6.png", "path_original": "poster_ocr/e12ac268034545ba993c226252692cd6.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['Trendy fashionconversation', 'join now']", "size": [ 1200, 1200 ], "texts": [ "Trendy fashionconversation", "join now" ], "text_bbox": [ [ 244, 370, 955, 512 ], [ 470, 169, 730, 201 ] ], "bbox_areas": [ 100962, 8320 ] }, { "task": "poster detection", "subtask": "", "name": "de0bc002ad1d4eaabe68ab86bdf554c2.png", "path": "poster_ocr_1024/de0bc002ad1d4eaabe68ab86bdf554c2.png", "path_original": "poster_ocr/de0bc002ad1d4eaabe68ab86bdf554c2.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['Blue', 'I prefer the white sky like cloudsSwim in the cloudsImagination is the most wonderful', 'Pure blue sky white clouds blossoming cloud like childhood innocence', 'sky Beautiful clouds ']", "size": [ 4000, 6000 ], "texts": [ "Blue", "I prefer the white sky like cloudsSwim in the cloudsImagination is the most wonderful", "Pure blue sky white clouds blossoming cloud like childhood innocence", "sky Beautiful clouds " ], "text_bbox": [ [ 926, 1237, 3348, 1941 ], [ 862, 5199, 3338, 5691 ], [ 819, 287, 3494, 544 ], [ 944, 1029, 3343, 1177 ] ], "bbox_areas": [ 1705088, 1218192, 687475, 355052 ] }, { "task": "poster detection", "subtask": "", "name": "dd5165533ee24a14b7b5e638df7086df.png", "path": "poster_ocr_1024/dd5165533ee24a14b7b5e638df7086df.png", "path_original": "poster_ocr/dd5165533ee24a14b7b5e638df7086df.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['POLYGON', 'MINIMAL DIEZ . ASTRO . ALPHA BEATZ', 'ELECTRO / HOUSE / DUBSTEP', 'YOURCLUB', '25/05/2015', 'www.yourwebsite.com']", "size": [ 1275, 1875 ], "texts": [ "POLYGON", "MINIMAL DIEZ . ASTRO . ALPHA BEATZ", "ELECTRO / HOUSE / DUBSTEP", "YOURCLUB", "25/05/2015", "www.yourwebsite.com" ], "text_bbox": [ [ 96, 1447, 679, 1542 ], [ 101, 1619, 769, 1648 ], [ 101, 1653, 669, 1685 ], [ 97, 1714, 437, 1761 ], [ 98, 1573, 351, 1614 ], [ 113, 1762, 389, 1791 ] ], "bbox_areas": [ 55385, 19372, 18176, 15980, 10373, 8004 ] }, { "task": "poster detection", "subtask": "", "name": "dca5bfcf26ca4701aa4f4e58341f3cbf.png", "path": "poster_ocr_1024/dca5bfcf26ca4701aa4f4e58341f3cbf.png", "path_original": "poster_ocr/dca5bfcf26ca4701aa4f4e58341f3cbf.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['MUSIC PARTY', 'CONCERT', 'ELECTRO', 'FUTURE SOUND PARTY FLYERBEST PARTY EVERS DRESS COAT ONLY REDCALL US FOR BOOKING MORE +45-123-456-456-789', 'Lorem Ipsum is simply dummy text of the printing and typesetting industry galley of type orem Ipsum is simply dummy text of the printing and typesetting industry galley of type.orem.Lorem Ipsum is simply dummy text of the printing and typesetting industry galley of type orem', 'JUNUARY25-12-2014-8PMBEST PARTY EVER WORLD', 'Lorem Ipsum is simply dummy text of the printing and typesetting industry galley of type.', 'SPECAIL GUESTDJ ALEXA | DJ JOHN', 'PRESENTANNAAYA-220']", "size": [ 1275, 1875 ], "texts": [ "MUSIC PARTY", "CONCERT", "ELECTRO", "FUTURE SOUND PARTY FLYERBEST PARTY EVERS DRESS COAT ONLY REDCALL US FOR BOOKING MORE +45-123-456-456-789", "Lorem Ipsum is simply dummy text of the printing and typesetting industry galley of type orem Ipsum is simply dummy text of the printing and typesetting industry galley of type.orem.Lorem Ipsum is simply dummy text of the printing and typesetting industry galley of type orem", "JUNUARY25-12-2014-8PMBEST PARTY EVER WORLD", "Lorem Ipsum is simply dummy text of the printing and typesetting industry galley of type.", "SPECAIL GUESTDJ ALEXA | DJ JOHN", "PRESENTANNAAYA-220" ], "text_bbox": [ [ 258, 718, 1006, 1035 ], [ 342, 1007, 851, 1259 ], [ 444, 497, 927, 744 ], [ 179, 1589, 1098, 1696 ], [ 114, 1718, 1156, 1797 ], [ 78, 78, 534, 184 ], [ 77, 199, 621, 249 ], [ 853, 160, 1191, 222 ], [ 912, 77, 1198, 146 ] ], "bbox_areas": [ 237116, 128268, 119301, 98333, 82318, 48336, 27200, 20956, 19734 ] }, { "task": "poster detection", "subtask": "", "name": "dc1371a5e2024b04be65eb6621c0cc32.png", "path": "poster_ocr_1024/dc1371a5e2024b04be65eb6621c0cc32.png", "path_original": "poster_ocr/dc1371a5e2024b04be65eb6621c0cc32.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['DRINKNIGHT', 'dj BLACK | Dj MONEY | DJ STRIKE', '-OVNI EVENT PRESENTs-', '9 PM . ENTRY 20$ . DRESS CODE : ALL IN BLACK', 'NightClub', 'FRIDAY 15th JULY', 'f facebook.com/YOURNAME', 'T twitter.com/YOURNAME']", "size": [ 1275, 1875 ], "texts": [ "DRINKNIGHT", "dj BLACK | Dj MONEY | DJ STRIKE", "-OVNI EVENT PRESENTs-", "9 PM . ENTRY 20$ . DRESS CODE : ALL IN BLACK", "NightClub", "FRIDAY 15th JULY", "f facebook.com/YOURNAME", "T twitter.com/YOURNAME" ], "text_bbox": [ [ 183, 1201, 1096, 1484 ], [ 184, 1565, 1089, 1624 ], [ 209, 1138, 1062, 1169 ], [ 189, 1517, 1085, 1546 ], [ 510, 1635, 778, 1696 ], [ 464, 1071, 806, 1117 ], [ 203, 1635, 506, 1667 ], [ 789, 1635, 1069, 1667 ] ], "bbox_areas": [ 258379, 53395, 26443, 25984, 16348, 15732, 9696, 8960 ] }, { "task": "poster detection", "subtask": "", "name": "dc258cc0b92c4827a4ea5f2b1eef62e3.png", "path": "poster_ocr_1024/dc258cc0b92c4827a4ea5f2b1eef62e3.png", "path_original": "poster_ocr/dc258cc0b92c4827a4ea5f2b1eef62e3.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['Fashionable and comfortable, our new line of atheltic wear will keep you sweat proof and looking great all day', 'get moving', 'shopnow']", "size": [ 750, 1050 ], "texts": [ "Fashionable and comfortable, our new line of atheltic wear will keep you sweat proof and looking great all day", "get moving", "shopnow" ], "text_bbox": [ [ 119, 209, 650, 298 ], [ 80, 60, 684, 126 ], [ 72, 629, 198, 717 ] ], "bbox_areas": [ 47259, 39864, 11088 ] }, { "task": "poster detection", "subtask": "", "name": "db05ba872b304dfb810b8640b76583b4.png", "path": "poster_ocr_1024/db05ba872b304dfb810b8640b76583b4.png", "path_original": "poster_ocr/db05ba872b304dfb810b8640b76583b4.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['SESSION', 'ROCK', 'FOR RESERVATIONS OR MORE INFO', 'SLAYMOORE', '14323MAIN STR.OLD FORTCA.', 'ROBSHOTS', 'THE OWLZ', 'INDEPENDENT MUSIC LIVE SHOW', 'AT THE INN', 'you are invited', 'doors open at 9PM', '25$ tickets 5$ drinks', 'FRIDAY', 'WWW.YOURWEBSITE.COM', 'SPECIAL GUEST', 'LIVE', '02.05.2014', 'presenting:', 'MUSIC', '222 333 555', 'or call us:', 'FACEBOOK', 'and', 'YOUTUBE', 'TWITTER']", "size": [ 2700, 3450 ], "texts": [ "SESSION", "ROCK", "FOR RESERVATIONS OR MORE INFO", "SLAYMOORE", "14323MAIN STR.OLD FORTCA.", "ROBSHOTS", "THE OWLZ", "INDEPENDENT MUSIC LIVE SHOW", "AT THE INN", "you are invited", "doors open at 9PM", "25$ tickets 5$ drinks", "FRIDAY", "WWW.YOURWEBSITE.COM", "SPECIAL GUEST", "LIVE", "02.05.2014", "presenting:", "MUSIC", "222 333 555", "or call us:", "FACEBOOK", "and", "YOUTUBE", "TWITTER" ], "text_bbox": [ [ 420, 879, 1723, 1078 ], [ 770, 645, 1724, 867 ], [ 508, 2889, 2184, 2968 ], [ 855, 1483, 1696, 1633 ], [ 1934, 1489, 2335, 1771 ], [ 760, 1634, 1497, 1781 ], [ 756, 1781, 1465, 1929 ], [ 342, 1089, 1721, 1160 ], [ 941, 1182, 1704, 1292 ], [ 1916, 675, 2000, 1427 ], [ 512, 2640, 1351, 2715 ], [ 506, 2739, 1349, 2805 ], [ 1179, 362, 1654, 471 ], [ 508, 2995, 1396, 3052 ], [ 916, 2061, 1487, 2142 ], [ 775, 2325, 1085, 2448 ], [ 1208, 500, 1620, 578 ], [ 1208, 1386, 1672, 1454 ], [ 779, 2457, 1083, 2541 ], [ 1824, 2997, 2184, 3052 ], [ 1466, 2998, 1759, 3045 ], [ 836, 3131, 1103, 3169 ], [ 1331, 1953, 1478, 2021 ], [ 1706, 3131, 1931, 3169 ], [ 1312, 3131, 1530, 3170 ] ], "bbox_areas": [ 259297, 211788, 132404, 126150, 113082, 108339, 104932, 97909, 83930, 63168, 62925, 55638, 51775, 50616, 46251, 38130, 32136, 31552, 25536, 19800, 13771, 10146, 9996, 8550, 8502 ] }, { "task": "poster detection", "subtask": "", "name": "da621b549b494495bd34a57ee2a1e7d6.png", "path": "poster_ocr_1024/da621b549b494495bd34a57ee2a1e7d6.png", "path_original": "poster_ocr/da621b549b494495bd34a57ee2a1e7d6.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['ARTISTIC', 'MUSIC', 'EVENT', '-YOUR EVENT PRESENTS-', 'DEEJAY MINIMAL . DEEJAY SOUND', 'www.yourwebsite.com', '19/06/2016', 'ELECTRO / HOUSE / DUBSTEP', 'YOURCLUB']", "size": [ 1275, 1875 ], "texts": [ "ARTISTIC", "MUSIC", "EVENT", "-YOUR EVENT PRESENTS-", "DEEJAY MINIMAL . DEEJAY SOUND", "www.yourwebsite.com", "19/06/2016", "ELECTRO / HOUSE / DUBSTEP", "YOURCLUB" ], "text_bbox": [ [ 432, 613, 837, 717 ], [ 490, 738, 780, 842 ], [ 534, 863, 741, 937 ], [ 349, 264, 907, 291 ], [ 411, 1533, 866, 1562 ], [ 429, 1689, 852, 1714 ], [ 516, 1476, 762, 1517 ], [ 430, 1571, 856, 1594 ], [ 544, 1632, 737, 1677 ] ], "bbox_areas": [ 42120, 30160, 15318, 15066, 13195, 10575, 10086, 9798, 8685 ] }, { "task": "poster detection", "subtask": "", "name": "da5d6e913fd943cab3b077160fec09fe.png", "path": "poster_ocr_1024/da5d6e913fd943cab3b077160fec09fe.png", "path_original": "poster_ocr/da5d6e913fd943cab3b077160fec09fe.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['Sale', 'Lorem ipsum dolor sit amet,consectetur adipiscing elit, sed doeiusmod tempor.', 'weekend', '25% till 75% off', 'midnight']", "size": [ 1000, 1000 ], "texts": [ "Sale", "Lorem ipsum dolor sit amet,consectetur adipiscing elit, sed doeiusmod tempor.", "weekend", "25% till 75% off", "midnight" ], "text_bbox": [ [ 35, 622, 365, 760 ], [ 46, 848, 355, 913 ], [ 72, 516, 325, 555 ], [ 63, 794, 336, 823 ], [ 114, 579, 285, 601 ] ], "bbox_areas": [ 45540, 20085, 9867, 7917, 3762 ] }, { "task": "poster detection", "subtask": "", "name": "d9db007d64e64d6199efc190d62abcd9.png", "path": "poster_ocr_1024/d9db007d64e64d6199efc190d62abcd9.png", "path_original": "poster_ocr/d9db007d64e64d6199efc190d62abcd9.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['Blues', 'J.Dan', 'Lorem ipsum dolor sit amet, consectetur adipiscing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua. Ut enim ad minim veniam, quis nostrud exercitation ullamco laboris nisi ut aliquip ex ea commodo consequat.', 'special guest', 'live music festival', 'best night concert', 'doors open at 9pm | free drinks | free merch | rsvp : 12345678', 'exclusiveflyer.net present', 'music festival present', 'Exclusive.FM', '$25', '11.PM', 'Drink and Parking', 'free', 'www.exclusiveflyer.net', 'facebook.com/party', 'twitter.com/party', 'Ticket']", "size": [ 1350, 1950 ], "texts": [ "Blues", "J.Dan", "Lorem ipsum dolor sit amet, consectetur adipiscing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua. Ut enim ad minim veniam, quis nostrud exercitation ullamco laboris nisi ut aliquip ex ea commodo consequat.", "special guest", "live music festival", "best night concert", "doors open at 9pm | free drinks | free merch | rsvp : 12345678", "exclusiveflyer.net present", "music festival present", "Exclusive.FM", "$25", "11.PM", "Drink and Parking", "free", "www.exclusiveflyer.net", "facebook.com/party", "twitter.com/party", "Ticket" ], "text_bbox": [ [ 47, 118, 1309, 490 ], [ 582, 1292, 1263, 1554 ], [ 70, 1413, 526, 1651 ], [ 585, 1566, 1254, 1661 ], [ 646, 1695, 1187, 1784 ], [ 579, 500, 1309, 551 ], [ 121, 1847, 1229, 1871 ], [ 149, 66, 1200, 84 ], [ 560, 1254, 1270, 1274 ], [ 185, 1365, 527, 1404 ], [ 38, 1255, 210, 1329 ], [ 307, 1671, 526, 1728 ], [ 190, 1778, 526, 1813 ], [ 385, 1735, 528, 1773 ], [ 538, 1884, 846, 1901 ], [ 216, 1884, 455, 1901 ], [ 915, 1885, 1134, 1901 ], [ 96, 1230, 206, 1257 ] ], "bbox_areas": [ 469464, 178422, 108528, 63555, 48149, 37230, 26592, 18918, 14200, 13338, 12728, 12483, 11760, 5434, 5236, 4063, 3504, 2970 ] }, { "task": "poster detection", "subtask": "", "name": "d829c902dba74012a90f885978fac7b3.png", "path": "poster_ocr_1024/d829c902dba74012a90f885978fac7b3.png", "path_original": "poster_ocr/d829c902dba74012a90f885978fac7b3.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['dressup!109looksdeluseexklusivebeauty-geschenkezu gewinnen', 'gonnacrazybydurasambientvol.1', 'd', 'la repubblica delle donne', 'beauty']", "size": [ 2100, 2850 ], "texts": [ "dressup!109looksdeluseexklusivebeauty-geschenkezu gewinnen", "gonnacrazybydurasambientvol.1", "d", "la repubblica delle donne", "beauty" ], "text_bbox": [ [ 55, 1102, 918, 2273 ], [ 1454, 2021, 2038, 2798 ], [ 81, 182, 672, 781 ], [ 142, 200, 1747, 339 ], [ 716, 393, 1664, 573 ] ], "bbox_areas": [ 1010573, 453768, 354009, 223095, 170640 ] }, { "task": "poster detection", "subtask": "", "name": "d7f4f7ee6b424b6fa42ca6e97c815fca.png", "path": "poster_ocr_1024/d7f4f7ee6b424b6fa42ca6e97c815fca.png", "path_original": "poster_ocr/d7f4f7ee6b424b6fa42ca6e97c815fca.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['WorldForest Day', 'Lorem ipsum dolor sit amet, consectetur adipiscing elit sed do eiusmod tempor incididunt ut labore et dolore.', 'Forest protection now', 'Lorem ipsum dolor sit amet, consectetur adipiscing elit sed do eiusmod tempor incididunt ut labore et dolore.', '21 March', 'www.yoursite.com', 'your account', 'your account', 'your account']", "size": [ 1753, 2480 ], "texts": [ "WorldForest Day", "Lorem ipsum dolor sit amet, consectetur adipiscing elit sed do eiusmod tempor incididunt ut labore et dolore.", "Forest protection now", "Lorem ipsum dolor sit amet, consectetur adipiscing elit sed do eiusmod tempor incididunt ut labore et dolore.", "21 March", "www.yoursite.com", "your account", "your account", "your account" ], "text_bbox": [ [ 742, 936, 1615, 1272 ], [ 860, 1424, 1611, 1680 ], [ 696, 1187, 1603, 1362 ], [ 539, 82, 1580, 176 ], [ 1241, 799, 1604, 869 ], [ 1137, 1782, 1545, 1807 ], [ 1000, 2309, 1218, 2341 ], [ 630, 2312, 848, 2344 ], [ 1363, 2306, 1581, 2338 ] ], "bbox_areas": [ 293328, 192256, 158725, 97854, 25410, 10200, 6976, 6976, 6976 ] }, { "task": "poster detection", "subtask": "", "name": "d787cedc87f64594a05a7e2de60478e4.png", "path": "poster_ocr_1024/d787cedc87f64594a05a7e2de60478e4.png", "path_original": "poster_ocr/d787cedc87f64594a05a7e2de60478e4.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['RETURN', 'OF THE 80S', 'FACEBOOK/YOURNAME TWITTER/YOURNAME YOUTUBE/YOURNAME', 'SAT. JUNE15TH', 'HOSTED BY MC DAVEY FEATURING DJ SLICK & DAZE', 'AT THE INN', '25$ ENTRY / FREE PARKING / DOORS OPEN AT 9PM', 'www.website.com', '021 main street city ca. near the plaza ']", "size": [ 2700, 3450 ], "texts": [ "RETURN", "OF THE 80S", "FACEBOOK/YOURNAME TWITTER/YOURNAME YOUTUBE/YOURNAME", "SAT. JUNE15TH", "HOSTED BY MC DAVEY FEATURING DJ SLICK & DAZE", "AT THE INN", "25$ ENTRY / FREE PARKING / DOORS OPEN AT 9PM", "www.website.com", "021 main street city ca. near the plaza " ], "text_bbox": [ [ 53, 73, 2152, 645 ], [ 500, 559, 2616, 964 ], [ 428, 1413, 2165, 1529 ], [ 1326, 1048, 2428, 1190 ], [ 185, 1202, 2320, 1269 ], [ 188, 1049, 1201, 1187 ], [ 183, 1281, 1874, 1346 ], [ 183, 1622, 1101, 1693 ], [ 183, 1346, 1031, 1380 ] ], "bbox_areas": [ 1200628, 856980, 201492, 156484, 143045, 139794, 109915, 65178, 28832 ] }, { "task": "poster detection", "subtask": "", "name": "d6cee2da43e446f5bc513a87da10c170.png", "path": "poster_ocr_1024/d6cee2da43e446f5bc513a87da10c170.png", "path_original": "poster_ocr/d6cee2da43e446f5bc513a87da10c170.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['OpenDoors', 'OpenDoors', 'dj FLYER | Dj LIQUID | DJ GRAPHICS', 'Club designers 123 St. YOUR CITY', 'SATURDAY 25th OCTOBER 2014', 'ENTRY 10$ free admission for ladies until 10', 'NightClub', '-YOUR ENTERTAINMENT PRESENTS-', 'F facebook.com/YOURNAME', 'T twitter.com/YOURNAME', 'www.clubname.com']", "size": [ 1275, 1875 ], "texts": [ "OpenDoors", "OpenDoors", "dj FLYER | Dj LIQUID | DJ GRAPHICS", "Club designers 123 St. YOUR CITY", "SATURDAY 25th OCTOBER 2014", "ENTRY 10$ free admission for ladies until 10", "NightClub", "-YOUR ENTERTAINMENT PRESENTS-", "F facebook.com/YOURNAME", "T twitter.com/YOURNAME", "www.clubname.com" ], "text_bbox": [ [ 161, 573, 1019, 1142 ], [ 161, 573, 1019, 1142 ], [ 256, 1456, 1023, 1527 ], [ 258, 1532, 1022, 1582 ], [ 333, 161, 944, 223 ], [ 258, 1589, 1019, 1617 ], [ 524, 1695, 764, 1764 ], [ 335, 120, 942, 146 ], [ 238, 1692, 515, 1719 ], [ 789, 1692, 1044, 1719 ], [ 540, 1761, 743, 1783 ] ], "bbox_areas": [ 488202, 488202, 54457, 38200, 37882, 21308, 16560, 15782, 7479, 6885, 4466 ] }, { "task": "poster detection", "subtask": "", "name": "d687ee7b1f4445d0b9ea1bc4719a0e18.png", "path": "poster_ocr_1024/d687ee7b1f4445d0b9ea1bc4719a0e18.png", "path_original": "poster_ocr/d687ee7b1f4445d0b9ea1bc4719a0e18.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['FASHIONselections', 'www.yourwebsite.com', 'CARLA ORTIZ . JENNA JIA . ELITE COUTURE', 'YOURCLUB', '26/06/2016', 'KARL GRANT PRESENTS']", "size": [ 1275, 1875 ], "texts": [ "FASHIONselections", "www.yourwebsite.com", "CARLA ORTIZ . JENNA JIA . ELITE COUTURE", "YOURCLUB", "26/06/2016", "KARL GRANT PRESENTS" ], "text_bbox": [ [ 82, 892, 1190, 1307 ], [ 189, 1764, 1189, 1802 ], [ 296, 1635, 1195, 1665 ], [ 789, 1554, 1197, 1616 ], [ 76, 118, 470, 179 ], [ 66, 78, 564, 106 ] ], "bbox_areas": [ 459820, 38000, 26970, 25296, 24034, 13944 ] }, { "task": "poster detection", "subtask": "", "name": "d58d478069f7414aaf6812dfed60887d.png", "path": "poster_ocr_1024/d58d478069f7414aaf6812dfed60887d.png", "path_original": "poster_ocr/d58d478069f7414aaf6812dfed60887d.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['RED CUP NIGHT', 'dj SKULL | Dj MONEY | DJ STRIKE', '9 PM || ENTRY 20$ || DRESS CODE : ALL IN BLACK', '-OVNI EVENT PRESENTs-', 'NightClub', 'FRIDAY 15th JULY', 'f facebook.com/YOURNAME', 'T twitter.com/YOURNAME']", "size": [ 1275, 1875 ], "texts": [ "RED CUP NIGHT", "dj SKULL | Dj MONEY | DJ STRIKE", "9 PM || ENTRY 20$ || DRESS CODE : ALL IN BLACK", "-OVNI EVENT PRESENTs-", "NightClub", "FRIDAY 15th JULY", "f facebook.com/YOURNAME", "T twitter.com/YOURNAME" ], "text_bbox": [ [ 167, 1158, 1105, 1398 ], [ 218, 1460, 1058, 1516 ], [ 235, 1538, 1044, 1573 ], [ 277, 1073, 991, 1099 ], [ 520, 1590, 770, 1649 ], [ 491, 1017, 777, 1056 ], [ 224, 1597, 509, 1627 ], [ 787, 1596, 1049, 1626 ] ], "bbox_areas": [ 225120, 47040, 28315, 18564, 14750, 11154, 8550, 7860 ] }, { "task": "poster detection", "subtask": "", "name": "d528841e9c164f98afd4935df1555932.png", "path": "poster_ocr_1024/d528841e9c164f98afd4935df1555932.png", "path_original": "poster_ocr/d528841e9c164f98afd4935df1555932.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['Classic', 'Party', 'Beach', 'ChillaxDeejays', 'MajesticCasual', 'Paulo Arruda', '15.09.2013', 'f v n ', 'u l w', 'Miami Beach', 'Ticket 20 $']", "size": [ 2555, 3583 ], "texts": [ "Classic", "Party", "Beach", "ChillaxDeejays", "MajesticCasual", "Paulo Arruda", "15.09.2013", "f v n ", "u l w", "Miami Beach", "Ticket 20 $" ], "text_bbox": [ [ 798, 1481, 1770, 1707 ], [ 941, 1968, 1648, 2229 ], [ 874, 1728, 1696, 1937 ], [ 111, 3073, 1175, 3224 ], [ 111, 2934, 1189, 3082 ], [ 113, 2786, 1197, 2914 ], [ 102, 101, 872, 215 ], [ 80, 3363, 1071, 3445 ], [ 1483, 3363, 2463, 3445 ], [ 1688, 3125, 2458, 3220 ], [ 1815, 116, 2436, 226 ] ], "bbox_areas": [ 219672, 184527, 171798, 160664, 159544, 138752, 87780, 81262, 80360, 73150, 68310 ] }, { "task": "poster detection", "subtask": "", "name": "d4f1411998424d16a0167b750f8d7ac7.png", "path": "poster_ocr_1024/d4f1411998424d16a0167b750f8d7ac7.png", "path_original": "poster_ocr/d4f1411998424d16a0167b750f8d7ac7.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['WEEKEND', 'SEP-15', 'MUSIC', 'YOURDJSHOW.COM', 'IF YOU WANT MORE INFO ABOUT THIS EVENT VISIT WWW.YOURDJSHOW.COM', 'OR IN THE NEWYORK STADIUM OFFICE', 'FREE ENTRY | REGISTER IN', '$10', '+21', 'ZACOMIC STUDIOS PRESENT', 'FREE COCKTAILS', 'UNTIL 10:00 PM', 'ENTRY']", "size": [ 1275, 1875 ], "texts": [ "WEEKEND", "SEP-15", "MUSIC", "YOURDJSHOW.COM", "IF YOU WANT MORE INFO ABOUT THIS EVENT VISIT WWW.YOURDJSHOW.COM", "OR IN THE NEWYORK STADIUM OFFICE", "FREE ENTRY | REGISTER IN", "$10", "+21", "ZACOMIC STUDIOS PRESENT", "FREE COCKTAILS", "UNTIL 10:00 PM", "ENTRY" ], "text_bbox": [ [ 124, 1177, 835, 1376 ], [ 685, 1467, 1132, 1650 ], [ 124, 1008, 475, 1160 ], [ 164, 1522, 660, 1596 ], [ 171, 1692, 1109, 1724 ], [ 152, 1612, 661, 1649 ], [ 305, 1465, 660, 1518 ], [ 841, 1007, 977, 1141 ], [ 1050, 677, 1186, 773 ], [ 490, 121, 783, 150 ], [ 977, 784, 1190, 820 ], [ 977, 829, 1180, 865 ], [ 751, 1058, 833, 1094 ] ], "bbox_areas": [ 141489, 81801, 53352, 36704, 30016, 18833, 18815, 18224, 13056, 8497, 7668, 7308, 2952 ] }, { "task": "poster detection", "subtask": "", "name": "d3bb12dda86746cba760b390eedd1bfe.png", "path": "poster_ocr_1024/d3bb12dda86746cba760b390eedd1bfe.png", "path_original": "poster_ocr/d3bb12dda86746cba760b390eedd1bfe.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['M', 'T', 'E', 'monster', \"monsterDon't leave me.\", 'May the joy and warmth ofChristmas fill your ', 'Long time no see.', 'I', \"Your hair The snow crumbling The dying birds singing in a lustrous voice The waves cried Your exuberant bluehands Bluethifted living with meon Hey you're so cool It's hard to keep \", 'may the joy and warmthof Christmas fill your home with happinessYou know what I used to look like time.', ' joy and happiness of Christmas', 'Wishing you the peace', \"don't leave \", 'Christmas']", "size": [ 3840, 5126 ], "texts": [ "M", "T", "E", "monster", "monsterDon't leave me.", "May the joy and warmth ofChristmas fill your ", "Long time no see.", "I", "Your hair The snow crumbling The dying birds singing in a lustrous voice The waves cried Your exuberant bluehands Bluethifted living with meon Hey you're so cool It's hard to keep ", "may the joy and warmthof Christmas fill your home with happinessYou know what I used to look like time.", " joy and happiness of Christmas", "Wishing you the peace", "don't leave ", "Christmas" ], "text_bbox": [ [ 1712, 2737, 2502, 3414 ], [ 275, 1865, 827, 2541 ], [ 2651, 4002, 3139, 4738 ], [ 384, 403, 1287, 614 ], [ 349, 1043, 1213, 1240 ], [ 171, 3591, 924, 3805 ], [ 319, 4005, 1042, 4226 ], [ 1010, 3356, 1195, 4212 ], [ 171, 4502, 1364, 4633 ], [ 483, 759, 1124, 857 ], [ 3654, 437, 3697, 1159 ], [ 3659, 1990, 3700, 2501 ], [ 632, 335, 1001, 387 ], [ 3658, 1514, 3693, 1744 ] ], "bbox_areas": [ 534830, 373152, 359168, 190533, 170208, 161142, 159783, 158360, 156283, 62818, 31046, 20951, 19188, 8050 ] }, { "task": "poster detection", "subtask": "", "name": "d31ff06da6864d308a2efe8d9b458e1f.png", "path": "poster_ocr_1024/d31ff06da6864d308a2efe8d9b458e1f.png", "path_original": "poster_ocr/d31ff06da6864d308a2efe8d9b458e1f.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['session', 'winter', 'minimal', 'dj cold i dj blue ', 'sat nov 15', 'free drinks i free for girls after 12 pm', 'club name i miami, fl 12345', 'featuring', 'awesomeflyer', 'doors open from 8 pm', 'presents']", "size": [ 1275, 1875 ], "texts": [ "session", "winter", "minimal", "dj cold i dj blue ", "sat nov 15", "free drinks i free for girls after 12 pm", "club name i miami, fl 12345", "featuring", "awesomeflyer", "doors open from 8 pm", "presents" ], "text_bbox": [ [ 258, 839, 1016, 972 ], [ 253, 681, 1022, 811 ], [ 255, 525, 1020, 652 ], [ 242, 1593, 1020, 1655 ], [ 404, 1462, 858, 1522 ], [ 276, 1718, 979, 1742 ], [ 367, 1755, 888, 1785 ], [ 382, 1543, 880, 1570 ], [ 451, 398, 823, 429 ], [ 423, 1674, 832, 1699 ], [ 531, 441, 743, 471 ] ], "bbox_areas": [ 100814, 99970, 97155, 48236, 27240, 16872, 15630, 13446, 11532, 10225, 6360 ] }, { "task": "poster detection", "subtask": "", "name": "d2ae7cef7d6c466ebcad674d4c42f4b6.png", "path": "poster_ocr_1024/d2ae7cef7d6c466ebcad674d4c42f4b6.png", "path_original": "poster_ocr/d2ae7cef7d6c466ebcad674d4c42f4b6.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['love', 'elec', 'tro', 'this autumnmost awaited event this yearbest deejaysmolestie a, ultricies porta urna. pat a, convallisfor more information:yourparty.comvelit, rhoncus eu, luctus et interdum adipiscing wisi. Aliquam erat ac ipsum. Integer aliquam purus. Quisque lorem tortor fringilla sed, vestibulum id, eleifend justo vel bibendum sapien massa ac turpis faucibus orci luctus non, consectetuer lobortis quis, varius in, purus. Integer ultrices posuere cubilia Curae, Nulla ipsum dolor lacus, suscipit adipiscing. Cum sociis natoque penatibus et ultrices volutpat. Nullam wisi ultricies a, gravida vitae, dapibus risus ante sodales lectus blandit eu, tempor diam pede cursus vitae, ultricies eu, faucibus quis, porttitor eros cursus lectus, pellentesque eget, bibendum a, gravida ullamcorper quam. Nullam viverra consectetuer. Quisque cursus et, porttitor risus. Aliquam sem. In hendrerit nulla quam nunc, accumsan congue. Lorem ipsum primis in nibh vel risus. Sed vel lectus. Ut sagittis, ipsum dolor quam.', '09.03', '22.11.201322:00CLUB NAMECITYGIRLS: 30$BOYS: 50$', 'Lorem ipsum pendisse a pelle', 'yourparty.com', '------------------------------------------------', '---------------------------------------------------------------------------------------', '------------------------------------', '------------------------']", "size": [ 2550, 3570 ], "texts": [ "love", "elec", "tro", "this autumnmost awaited event this yearbest deejaysmolestie a, ultricies porta urna. pat a, convallisfor more information:yourparty.comvelit, rhoncus eu, luctus et interdum adipiscing wisi. Aliquam erat ac ipsum. Integer aliquam purus. Quisque lorem tortor fringilla sed, vestibulum id, eleifend justo vel bibendum sapien massa ac turpis faucibus orci luctus non, consectetuer lobortis quis, varius in, purus. Integer ultrices posuere cubilia Curae, Nulla ipsum dolor lacus, suscipit adipiscing. Cum sociis natoque penatibus et ultrices volutpat. Nullam wisi ultricies a, gravida vitae, dapibus risus ante sodales lectus blandit eu, tempor diam pede cursus vitae, ultricies eu, faucibus quis, porttitor eros cursus lectus, pellentesque eget, bibendum a, gravida ullamcorper quam. Nullam viverra consectetuer. Quisque cursus et, porttitor risus. Aliquam sem. In hendrerit nulla quam nunc, accumsan congue. Lorem ipsum primis in nibh vel risus. Sed vel lectus. Ut sagittis, ipsum dolor quam.", "09.03", "22.11.201322:00CLUB NAMECITYGIRLS: 30$BOYS: 50$", "Lorem ipsum pendisse a pelle", "yourparty.com", "------------------------------------------------", "---------------------------------------------------------------------------------------", "------------------------------------", "------------------------" ], "text_bbox": [ [ 399, 358, 1409, 700 ], [ 396, 733, 1295, 1075 ], [ 411, 1110, 1120, 1451 ], [ 272, 2367, 1019, 2690 ], [ 265, 1905, 786, 2094 ], [ 2066, 255, 2307, 581 ], [ 270, 2223, 953, 2254 ], [ 282, 2770, 648, 2801 ], [ 784, 1553, 786, 2293 ], [ 0, 356, 707, 357 ], [ 1555, 359, 2109, 360 ], [ 1279, 1451, 1647, 1452 ] ], "bbox_areas": [ 345420, 307458, 241769, 241281, 98469, 78566, 21173, 11346, 1480, 707, 554, 368 ] }, { "task": "poster detection", "subtask": "", "name": "d251909b6e254be7a630fb3552db2ea5.png", "path": "poster_ocr_1024/d251909b6e254be7a630fb3552db2ea5.png", "path_original": "poster_ocr/d251909b6e254be7a630fb3552db2ea5.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['harmony', 'RiverSide Bar', 'DJ SEVEN DJ STYLES DJ ENVATO', 'SATURDAY JULY 27', 'DOORS OPEN AT 9PM DRESS TO IMPRESS', 'SEVENSTYLES PRESENTS', 'WWW.YOURWEBSITE.COM', 'MUSIC PERFORMED BY']", "size": [ 1275, 1875 ], "texts": [ "harmony", "RiverSide Bar", "DJ SEVEN DJ STYLES DJ ENVATO", "SATURDAY JULY 27", "DOORS OPEN AT 9PM DRESS TO IMPRESS", "SEVENSTYLES PRESENTS", "WWW.YOURWEBSITE.COM", "MUSIC PERFORMED BY" ], "text_bbox": [ [ 259, 1058, 1045, 1421 ], [ 338, 1667, 942, 1774 ], [ 295, 1495, 974, 1518 ], [ 412, 1310, 863, 1340 ], [ 343, 1610, 927, 1628 ], [ 462, 58, 813, 76 ], [ 455, 1804, 820, 1821 ], [ 476, 1452, 795, 1469 ] ], "bbox_areas": [ 285318, 64628, 15617, 13530, 10512, 6318, 6205, 5423 ] }, { "task": "poster detection", "subtask": "", "name": "d1dd324043004842a3b784ab204fa515.png", "path": "poster_ocr_1024/d1dd324043004842a3b784ab204fa515.png", "path_original": "poster_ocr/d1dd324043004842a3b784ab204fa515.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['for fall', 'shop137162404.taobao.com', 'best skinny jeans']", "size": [ 750, 1050 ], "texts": [ "for fall", "shop137162404.taobao.com", "best skinny jeans" ], "text_bbox": [ [ 91, 556, 671, 709 ], [ 100, 985, 627, 1007 ], [ 194, 500, 557, 521 ] ], "bbox_areas": [ 88740, 11594, 7623 ] }, { "task": "poster detection", "subtask": "", "name": "d0e5dea1f98f41d9a403fe45fa5b5cfa.png", "path": "poster_ocr_1024/d0e5dea1f98f41d9a403fe45fa5b5cfa.png", "path_original": "poster_ocr/d0e5dea1f98f41d9a403fe45fa5b5cfa.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['hottlist', 'Clothes we are crushing over for New York fashion week', 'The']", "size": [ 750, 1050 ], "texts": [ "hottlist", "Clothes we are crushing over for New York fashion week", "The" ], "text_bbox": [ [ 105, 350, 645, 651 ], [ 115, 762, 632, 839 ], [ 322, 259, 425, 315 ] ], "bbox_areas": [ 162540, 39809, 5768 ] }, { "task": "poster detection", "subtask": "", "name": "d05d4db31605467fbfd7f910ebba7163.png", "path": "poster_ocr_1024/d05d4db31605467fbfd7f910ebba7163.png", "path_original": "poster_ocr/d05d4db31605467fbfd7f910ebba7163.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['20', 'Why wait? Get Started Today ', 'Be quick! Sale Ends on Sunday!', '% off', 'USE THE CODE : SPRING2017', '3 DAYS LEFT']", "size": [ 1080, 1080 ], "texts": [ "20", "Why wait? Get Started Today ", "Be quick! Sale Ends on Sunday!", "% off", "USE THE CODE : SPRING2017", "3 DAYS LEFT" ], "text_bbox": [ [ 267, 493, 590, 725 ], [ 179, 770, 901, 838 ], [ 281, 855, 822, 901 ], [ 625, 506, 799, 583 ], [ 303, 978, 786, 997 ], [ 612, 631, 813, 650 ] ], "bbox_areas": [ 74936, 49096, 24886, 13398, 9177, 3819 ] }, { "task": "poster detection", "subtask": "", "name": "ceb107f8d8234bfa94669159da899e58.png", "path": "poster_ocr_1024/ceb107f8d8234bfa94669159da899e58.png", "path_original": "poster_ocr/ceb107f8d8234bfa94669159da899e58.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['Lemon baby', 'Sunshine Smile']", "size": [ 4000, 6000 ], "texts": [ "Lemon baby", "Sunshine Smile" ], "text_bbox": [ [ 485, 452, 3566, 938 ], [ 629, 875, 2122, 1043 ] ], "bbox_areas": [ 1497366, 250824 ] }, { "task": "poster detection", "subtask": "", "name": "ce975ffbfd474b4390c7d5369809e6ce.png", "path": "poster_ocr_1024/ce975ffbfd474b4390c7d5369809e6ce.png", "path_original": "poster_ocr/ce975ffbfd474b4390c7d5369809e6ce.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['$10Entry', '$10Entry', '$10Entry', 'Right on monday to weddayloremian grass-fed beef tenderldin', 'Right on monday to weddayloremian grass-fed beef tenderldin', 'Right on monday to weddayloremian grass-fed beef tenderldin', 'A Biggest Event', 'A Biggest Event', 'A Biggest Event', 'rerved with atparagus, Troccoli & Newotatoes', 'rerved with atparagus, Troccoli & Newotatoes', 'rerved with atparagus, Troccoli & Newotatoes', 'Call us :+098 123 4556 234', 'Call us :+098 123 4556 234', 'Call us :+098 123 4556 234']", "size": [ 1275, 1875 ], "texts": [ "$10Entry", "$10Entry", "$10Entry", "Right on monday to weddayloremian grass-fed beef tenderldin", "Right on monday to weddayloremian grass-fed beef tenderldin", "Right on monday to weddayloremian grass-fed beef tenderldin", "A Biggest Event", "A Biggest Event", "A Biggest Event", "rerved with atparagus, Troccoli & Newotatoes", "rerved with atparagus, Troccoli & Newotatoes", "rerved with atparagus, Troccoli & Newotatoes", "Call us :+098 123 4556 234", "Call us :+098 123 4556 234", "Call us :+098 123 4556 234" ], "text_bbox": [ [ 288, 521, 991, 1031 ], [ 288, 521, 991, 1031 ], [ 288, 521, 991, 1031 ], [ 153, 1564, 1131, 1664 ], [ 153, 1564, 1131, 1664 ], [ 153, 1564, 1131, 1664 ], [ 364, 1463, 902, 1534 ], [ 364, 1463, 902, 1534 ], [ 364, 1463, 902, 1534 ], [ 188, 1688, 1088, 1729 ], [ 188, 1688, 1088, 1729 ], [ 188, 1688, 1088, 1729 ], [ 106, 102, 482, 187 ], [ 106, 102, 482, 187 ], [ 106, 102, 482, 187 ] ], "bbox_areas": [ 358530, 358530, 358530, 97800, 97800, 97800, 38198, 38198, 38198, 36900, 36900, 36900, 31960, 31960, 31960 ] }, { "task": "poster detection", "subtask": "", "name": "cd67ca69376c437996d0bfab79c1c523.png", "path": "poster_ocr_1024/cd67ca69376c437996d0bfab79c1c523.png", "path_original": "poster_ocr/cd67ca69376c437996d0bfab79c1c523.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['Heritage', 'For the modern Style, sophisticatedand sartorially. Cuts andneutral palettes.', 'Lorem ipsum dolor sit amet, consectetur adipiscing elit,sed do eiusmod tempor.', 'For the Classic Vintage StyleAmericana meets old fashion boys.Subtle statements.', 'Minimalist', 'New Fashion']", "size": [ 1000, 1000 ], "texts": [ "Heritage", "For the modern Style, sophisticatedand sartorially. Cuts andneutral palettes.", "Lorem ipsum dolor sit amet, consectetur adipiscing elit,sed do eiusmod tempor.", "For the Classic Vintage StyleAmericana meets old fashion boys.Subtle statements.", "Minimalist", "New Fashion" ], "text_bbox": [ [ 610, 230, 905, 331 ], [ 96, 836, 439, 910 ], [ 224, 83, 777, 126 ], [ 589, 375, 926, 444 ], [ 113, 684, 422, 750 ], [ 363, 28, 635, 61 ] ], "bbox_areas": [ 29795, 25382, 23779, 23253, 20394, 8976 ] }, { "task": "poster detection", "subtask": "", "name": "cd3a0d94b789437f826438fdffac91f8.png", "path": "poster_ocr_1024/cd3a0d94b789437f826438fdffac91f8.png", "path_original": "poster_ocr/cd3a0d94b789437f826438fdffac91f8.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['ABSTRACT', 'MUSIC', '19/06/2016', 'DEEJAY MINIMAL . DEEJAY SOUND', 'EVENT', 'ELECTRO / HOUSE / DUBSTEP', 'YOURCLUB', 'www.yourwebsite.com']", "size": [ 1275, 1875 ], "texts": [ "ABSTRACT", "MUSIC", "19/06/2016", "DEEJAY MINIMAL . DEEJAY SOUND", "EVENT", "ELECTRO / HOUSE / DUBSTEP", "YOURCLUB", "www.yourwebsite.com" ], "text_bbox": [ [ 416, 478, 864, 582 ], [ 490, 603, 780, 707 ], [ 449, 1304, 828, 1365 ], [ 379, 1389, 898, 1423 ], [ 534, 728, 741, 802 ], [ 399, 1432, 886, 1458 ], [ 541, 1494, 734, 1539 ], [ 510, 1548, 759, 1572 ] ], "bbox_areas": [ 46592, 30160, 23119, 17646, 15318, 12662, 8685, 5976 ] }, { "task": "poster detection", "subtask": "", "name": "cbaf73f0eee04a1e834b33e2da89cefd.png", "path": "poster_ocr_1024/cbaf73f0eee04a1e834b33e2da89cefd.png", "path_original": "poster_ocr/cbaf73f0eee04a1e834b33e2da89cefd.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['We,TongRo Image Stock, since commence with producing digital Image slidetransparency business in 1992, have been building outstandingsuccess in distributing various kind of collections from overseas countriesto Korea and in supplying our own collectionsto many countries through around ', 'tongro hotelgangnam-ku, seoul, korea', 'brandlaunching party', \"I'd like to invite you to a party\", 'march 17, 06:00PM ', 'place', 'date']", "size": [ 3543, 4961 ], "texts": [ "We,TongRo Image Stock, since commence with producing digital Image slidetransparency business in 1992, have been building outstandingsuccess in distributing various kind of collections from overseas countriesto Korea and in supplying our own collectionsto many countries through around ", "tongro hotelgangnam-ku, seoul, korea", "brandlaunching party", "I'd like to invite you to a party", "march 17, 06:00PM ", "place", "date" ], "text_bbox": [ [ 1222, 3683, 2200, 3914 ], [ 1239, 3120, 2193, 3249 ], [ 1328, 1448, 2110, 1595 ], [ 1222, 3593, 2207, 3638 ], [ 1334, 2726, 2066, 2786 ], [ 1578, 2987, 1855, 3050 ], [ 1603, 2615, 1829, 2676 ] ], "bbox_areas": [ 225918, 123066, 114954, 44325, 43920, 17451, 13786 ] }, { "task": "poster detection", "subtask": "", "name": "cb6cb84e49764c19af6f57441728cb85.png", "path": "poster_ocr_1024/cb6cb84e49764c19af6f57441728cb85.png", "path_original": "poster_ocr/cb6cb84e49764c19af6f57441728cb85.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['electrosessions.', 'Saturday, 10 April 2013', 'At Future Stage Arena - Amsterdam', 'THEBESTCLUB PRESENTS', '8pm - 5am I $10 entry', 'electro - trance - hardstyle - and more', 'For more information go to www.thebestclub.com | thebest@club.com', 'PLEASE DRINK RESPONSABLY AND ENJOY THE MUSIC', 'celldweller,', ',emma hewitt ', ',dash berlin', 'guest mix,']", "size": [ 1890, 2598 ], "texts": [ "electrosessions.", "Saturday, 10 April 2013", "At Future Stage Arena - Amsterdam", "THEBESTCLUB PRESENTS", "8pm - 5am I $10 entry", "electro - trance - hardstyle - and more", "For more information go to www.thebestclub.com | thebest@club.com", "PLEASE DRINK RESPONSABLY AND ENJOY THE MUSIC", "celldweller,", ",emma hewitt ", ",dash berlin", "guest mix," ], "text_bbox": [ [ 292, 1090, 1343, 1594 ], [ 665, 1626, 1220, 1862 ], [ 409, 2262, 1455, 2339 ], [ 746, 131, 1168, 234 ], [ 645, 2115, 1284, 2183 ], [ 633, 2026, 1384, 2081 ], [ 442, 2447, 1387, 2478 ], [ 553, 2402, 1325, 2428 ], [ 1041, 852, 1493, 895 ], [ 444, 719, 887, 760 ], [ 333, 950, 739, 993 ], [ 1156, 946, 1449, 983 ] ], "bbox_areas": [ 529704, 130980, 80542, 43466, 43452, 41305, 29295, 20072, 19436, 18163, 17458, 10841 ] }, { "task": "poster detection", "subtask": "", "name": "cb2896c9b7844869a141996afda02208.png", "path": "poster_ocr_1024/cb2896c9b7844869a141996afda02208.png", "path_original": "poster_ocr/cb2896c9b7844869a141996afda02208.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['ELECTRO', 'ELECTRO', 'NIGHTS', 'NIGHTS', 'MORE TEXT HERE OR DETAILS IF NECESSSARY. SPONOSORS, DETAILS ETC. DELETE THE TEXT AND USE YOUR LOGO HERE. MRE TEXT OR DETAILS IF NECESSARY, SPONSORS, INFO.', '@ CLUB ZEPPELIN', 'DJSLICK', '15TH', 'JUNE', 'DOORS OPEN AT 9PM MAIN STREET CITY CA. ', '25$ENTRY>>>', 'SATURDAY', '>>>FEATURING', 'WWW.WEBSITE.COM ', 'ZEPPELIN PRESENTS']", "size": [ 2700, 3450 ], "texts": [ "ELECTRO", "ELECTRO", "NIGHTS", "NIGHTS", "MORE TEXT HERE OR DETAILS IF NECESSSARY. SPONOSORS, DETAILS ETC. DELETE THE TEXT AND USE YOUR LOGO HERE. MRE TEXT OR DETAILS IF NECESSARY, SPONSORS, INFO.", "@ CLUB ZEPPELIN", "DJSLICK", "15TH", "JUNE", "DOORS OPEN AT 9PM MAIN STREET CITY CA. ", "25$ENTRY>>>", "SATURDAY", ">>>FEATURING", "WWW.WEBSITE.COM ", "ZEPPELIN PRESENTS" ], "text_bbox": [ [ 194, 1629, 2213, 1959 ], [ 194, 1599, 2213, 1929 ], [ 194, 1977, 1816, 2312 ], [ 194, 1947, 1816, 2282 ], [ 193, 3098, 1912, 3175 ], [ 191, 2885, 1138, 2978 ], [ 191, 2686, 808, 2804 ], [ 247, 564, 670, 726 ], [ 247, 404, 670, 554 ], [ 193, 2992, 1504, 3036 ], [ 194, 848, 410, 1045 ], [ 247, 331, 670, 395 ], [ 191, 2584, 462, 2671 ], [ 191, 3260, 731, 3301 ], [ 191, 169, 586, 201 ] ], "bbox_areas": [ 666270, 666270, 543370, 543370, 132363, 88071, 72806, 68526, 63450, 57684, 42552, 27072, 23577, 22140, 12640 ] }, { "task": "poster detection", "subtask": "", "name": "cacb132dff2f427b8f7dbbf7be393c21.png", "path": "poster_ocr_1024/cacb132dff2f427b8f7dbbf7be393c21.png", "path_original": "poster_ocr/cacb132dff2f427b8f7dbbf7be393c21.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['LUXURY', 'LUXURY', 'Night', 'Night', 'DOORS OPEN: 9:00PMADMISSION: $10 | FREE DRINKS AT 11FOR RESERVATIONS CALL: (+597) 850 85469www.yourclub.com', 'Your Club', 'music bY: dj honey | dj white | dj gold', 'Saturday04.aug', 'PRESENTS', 'Faceboob.com/YOURCLUB', 'twitter.com/YOURCLUB']", "size": [ 1275, 1875 ], "texts": [ "LUXURY", "LUXURY", "Night", "Night", "DOORS OPEN: 9:00PMADMISSION: $10 | FREE DRINKS AT 11FOR RESERVATIONS CALL: (+597) 850 85469www.yourclub.com", "Your Club", "music bY: dj honey | dj white | dj gold", "Saturday04.aug", "PRESENTS", "Faceboob.com/YOURCLUB", "twitter.com/YOURCLUB" ], "text_bbox": [ [ 128, 716, 1152, 1060 ], [ 132, 716, 1156, 1060 ], [ 435, 984, 1167, 1311 ], [ 435, 984, 1167, 1311 ], [ 304, 1580, 965, 1731 ], [ 365, 73, 727, 164 ], [ 276, 1738, 993, 1780 ], [ 433, 1517, 845, 1564 ], [ 441, 159, 801, 194 ], [ 397, 1789, 629, 1809 ], [ 702, 1789, 921, 1809 ] ], "bbox_areas": [ 352256, 352256, 239364, 239364, 99811, 32942, 30114, 19364, 12600, 4640, 4380 ] }, { "task": "poster detection", "subtask": "", "name": "ca72a8bf88534c3581a5beb93a4b4737.png", "path": "poster_ocr_1024/ca72a8bf88534c3581a5beb93a4b4737.png", "path_original": "poster_ocr/ca72a8bf88534c3581a5beb93a4b4737.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['Back in season ', 'Double up on layers and texture to make a subtle,laidback, everyday look.', 'shop my favorite picks for the season', 'style tip', 'signature style']", "size": [ 750, 1050 ], "texts": [ "Back in season ", "Double up on layers and texture to make a subtle,laidback, everyday look.", "shop my favorite picks for the season", "style tip", "signature style" ], "text_bbox": [ [ 49, 160, 704, 218 ], [ 77, 624, 354, 732 ], [ 107, 249, 643, 286 ], [ 95, 554, 324, 594 ], [ 206, 100, 545, 124 ] ], "bbox_areas": [ 37990, 29916, 19832, 9160, 8136 ] }, { "task": "poster detection", "subtask": "", "name": "c9fa954e283f4ec8a4eb66ae8a8d4f56.png", "path": "poster_ocr_1024/c9fa954e283f4ec8a4eb66ae8a8d4f56.png", "path_original": "poster_ocr/c9fa954e283f4ec8a4eb66ae8a8d4f56.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['TravelerStories', \"Traveler's help\", 'Unforgettableand beautifulGreece', 'www.creatogallery.com', 'Vizit My Blog']", "size": [ 1200, 1200 ], "texts": [ "TravelerStories", "Traveler's help", "Unforgettableand beautifulGreece", "www.creatogallery.com", "Vizit My Blog" ], "text_bbox": [ [ 247, 382, 758, 606 ], [ 248, 299, 502, 376 ], [ 326, 687, 488, 756 ], [ 100, 1069, 270, 1083 ], [ 499, 1069, 623, 1087 ] ], "bbox_areas": [ 114464, 19558, 11178, 2380, 2232 ] }, { "task": "poster detection", "subtask": "", "name": "c800aa6b6b684bf4924ce9773107b167.png", "path": "poster_ocr_1024/c800aa6b6b684bf4924ce9773107b167.png", "path_original": "poster_ocr/c800aa6b6b684bf4924ce9773107b167.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['NIGHTDRINK', 'dj BLACK | Dj MONEY | DJ STRIKE', '9 PM || ENTRY 20$ || DRESS CODE : ALL IN BLACK', '-OVNI EVENT PRESENTs-', 'YOURNAME', 'YOURNAME', 'YOURNAME', 'NightClub', 'FRIDAY 15th JULY', 'f facebook.com/YOURNAME', 'T twitter.com/YOURNAME']", "size": [ 1275, 1875 ], "texts": [ "NIGHTDRINK", "dj BLACK | Dj MONEY | DJ STRIKE", "9 PM || ENTRY 20$ || DRESS CODE : ALL IN BLACK", "-OVNI EVENT PRESENTs-", "YOURNAME", "YOURNAME", "YOURNAME", "NightClub", "FRIDAY 15th JULY", "f facebook.com/YOURNAME", "T twitter.com/YOURNAME" ], "text_bbox": [ [ 173, 243, 1103, 563 ], [ 227, 1498, 1047, 1560 ], [ 195, 1582, 1081, 1620 ], [ 210, 155, 1063, 186 ], [ 534, 841, 757, 959 ], [ 534, 841, 757, 959 ], [ 534, 841, 757, 959 ], [ 510, 1639, 778, 1700 ], [ 465, 88, 807, 134 ], [ 203, 1639, 506, 1671 ], [ 789, 1639, 1069, 1671 ] ], "bbox_areas": [ 297600, 50840, 33668, 26443, 26314, 26314, 26314, 16348, 15732, 9696, 8960 ] }, { "task": "poster detection", "subtask": "", "name": "c77d4dab64fa42d785f41e6b912e4dc8.png", "path": "poster_ocr_1024/c77d4dab64fa42d785f41e6b912e4dc8.png", "path_original": "poster_ocr/c77d4dab64fa42d785f41e6b912e4dc8.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['BE BEAUTY IN NATURAL WAY', 'Julie Anderson', '30% FOR THE FIRST 10 BOOKED', 'presents']", "size": [ 1080, 1080 ], "texts": [ "BE BEAUTY IN NATURAL WAY", "Julie Anderson", "30% FOR THE FIRST 10 BOOKED", "presents" ], "text_bbox": [ [ 218, 779, 856, 911 ], [ 336, 82, 741, 140 ], [ 279, 977, 802, 995 ], [ 477, 148, 606, 178 ] ], "bbox_areas": [ 84216, 23490, 9414, 3870 ] }, { "task": "poster detection", "subtask": "", "name": "c6e0d4af1aad4365a8969a658bcd7c0c.png", "path": "poster_ocr_1024/c6e0d4af1aad4365a8969a658bcd7c0c.png", "path_original": "poster_ocr/c6e0d4af1aad4365a8969a658bcd7c0c.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['WHITE', 'NIGHTS', 'DJ SLICK | DJ PAUL | MC DAVEY', 'SATURDAY AUGUST 21ST', '10$ DRINKS | FREE PARKING | DOORS OPEN AT 9PM', 'SUMMERFEST', '2014 EDITION', '@CLUB ZEPPELIN', '7654 MAIN STREET, CITY CA. NEAR THE PLAZA', 'ENTRY', 'FREE', 'ZEPPELIN PRESENTS', 'ALL NIGHT', 'FACEBOOK', 'YOUTUBE', 'TWITTER', 'VIMEO']", "size": [ 2700, 3450 ], "texts": [ "WHITE", "NIGHTS", "DJ SLICK | DJ PAUL | MC DAVEY", "SATURDAY AUGUST 21ST", "10$ DRINKS | FREE PARKING | DOORS OPEN AT 9PM", "SUMMERFEST", "2014 EDITION", "@CLUB ZEPPELIN", "7654 MAIN STREET, CITY CA. NEAR THE PLAZA", "ENTRY", "FREE", "ZEPPELIN PRESENTS", "ALL NIGHT", "FACEBOOK", "YOUTUBE", "TWITTER", "VIMEO" ], "text_bbox": [ [ 496, 1092, 2177, 1630 ], [ 662, 1552, 2081, 1965 ], [ 638, 2790, 2060, 2918 ], [ 1016, 2034, 1709, 2274 ], [ 512, 2925, 2186, 3019 ], [ 899, 359, 1806, 465 ], [ 1019, 917, 1691, 1054 ], [ 898, 2577, 1805, 2671 ], [ 733, 3129, 1966, 3180 ], [ 380, 499, 764, 622 ], [ 440, 394, 697, 507 ], [ 1077, 122, 1637, 167 ], [ 408, 613, 733, 686 ], [ 429, 3277, 697, 3315 ], [ 2082, 3275, 2319, 3313 ], [ 989, 3279, 1206, 3315 ], [ 1555, 3277, 1720, 3315 ] ], "bbox_areas": [ 904378, 586047, 182016, 166320, 157356, 96142, 92064, 85258, 62883, 47232, 29041, 25200, 23725, 10184, 9006, 7812, 6270 ] }, { "task": "poster detection", "subtask": "", "name": "c66c41141bf14a36b719da0a42bb8d1d.png", "path": "poster_ocr_1024/c66c41141bf14a36b719da0a42bb8d1d.png", "path_original": "poster_ocr/c66c41141bf14a36b719da0a42bb8d1d.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['BISTRO', '31.03.14', 'NIGHT CLUB', 'IF YOU WANT MORE INFO VISTI OUR WEB PAGE WWW.OURSHOW.COM', 'GRAND STADIUM', 'FREE ENTRY TO ALL WOMEN', 'LIVE MUSIC UNTIL 23HRS']", "size": [ 1275, 1875 ], "texts": [ "BISTRO", "31.03.14", "NIGHT CLUB", "IF YOU WANT MORE INFO VISTI OUR WEB PAGE WWW.OURSHOW.COM", "GRAND STADIUM", "FREE ENTRY TO ALL WOMEN", "LIVE MUSIC UNTIL 23HRS" ], "text_bbox": [ [ 196, 1095, 1071, 1281 ], [ 256, 1527, 653, 1661 ], [ 315, 1342, 967, 1400 ], [ 258, 1690, 1015, 1721 ], [ 693, 1568, 1017, 1622 ], [ 696, 1637, 962, 1663 ], [ 692, 1528, 936, 1555 ] ], "bbox_areas": [ 162750, 53198, 37816, 23467, 17496, 6916, 6588 ] }, { "task": "poster detection", "subtask": "", "name": "c6305bf2f5794e3288dc6ea99e03572f.png", "path": "poster_ocr_1024/c6305bf2f5794e3288dc6ea99e03572f.png", "path_original": "poster_ocr/c6305bf2f5794e3288dc6ea99e03572f.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['ALONE', 'DJ ASTRO | LUKE WALKER | LIZ MARS ', '9 PM . ENTRY 20$ . DRESS CODE : GALACTIC STYLE', 'WWW.SPACESHIP.COM', '- SPACESHIP PRESENTS - ', '15.11.15', 'SpacePlace', 'f facebook.com/YOURNAME', 'T twitter.com/YOURNAME']", "size": [ 1275, 1875 ], "texts": [ "ALONE", "DJ ASTRO | LUKE WALKER | LIZ MARS ", "9 PM . ENTRY 20$ . DRESS CODE : GALACTIC STYLE", "WWW.SPACESHIP.COM", "- SPACESHIP PRESENTS - ", "15.11.15", "SpacePlace", "f facebook.com/YOURNAME", "T twitter.com/YOURNAME" ], "text_bbox": [ [ 251, 239, 1022, 418 ], [ 222, 1507, 1056, 1540 ], [ 227, 1553, 1052, 1582 ], [ 188, 1743, 1087, 1763 ], [ 236, 114, 987, 134 ], [ 498, 1437, 785, 1484 ], [ 534, 1604, 762, 1650 ], [ 271, 1614, 525, 1639 ], [ 770, 1613, 1004, 1638 ] ], "bbox_areas": [ 138009, 27522, 23925, 17980, 15020, 13489, 10488, 6350, 5850 ] }, { "task": "poster detection", "subtask": "", "name": "c5a5a97df43143fd97b107dbd0f8e731.png", "path": "poster_ocr_1024/c5a5a97df43143fd97b107dbd0f8e731.png", "path_original": "poster_ocr/c5a5a97df43143fd97b107dbd0f8e731.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['THE LUXURY', 'dj LUXE | Dj MONEY | DJ CLASSY', '9 PM || ENTRY 20$ || DRESS CODE : ALL RICHES', '-OVNI EVENT PRESENTs-', 'FRIDAY 25th DECEMBER', 'NightClub', 'f facebook.com/YOURNAME', 'YourName', 'T twitter.com/YOURNAME']", "size": [ 1275, 1875 ], "texts": [ "THE LUXURY", "dj LUXE | Dj MONEY | DJ CLASSY", "9 PM || ENTRY 20$ || DRESS CODE : ALL RICHES", "-OVNI EVENT PRESENTs-", "FRIDAY 25th DECEMBER", "NightClub", "f facebook.com/YOURNAME", "YourName", "T twitter.com/YOURNAME" ], "text_bbox": [ [ 227, 1243, 1057, 1507 ], [ 210, 1521, 1068, 1586 ], [ 222, 1599, 1045, 1627 ], [ 277, 1190, 991, 1216 ], [ 436, 1134, 830, 1173 ], [ 525, 1646, 762, 1701 ], [ 202, 1646, 514, 1675 ], [ 614, 814, 708, 909 ], [ 780, 1646, 1069, 1675 ] ], "bbox_areas": [ 219120, 55770, 23044, 18564, 15366, 13035, 9048, 8930, 8381 ] }, { "task": "poster detection", "subtask": "", "name": "c557f7c4497f4bd5a276db6aa3ca4e62.png", "path": "poster_ocr_1024/c557f7c4497f4bd5a276db6aa3ca4e62.png", "path_original": "poster_ocr/c557f7c4497f4bd5a276db6aa3ca4e62.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['ANCIENT.FUTURE', 'SATURDAY JUNE 30th', 'ELECTRO / HOUSE / DUBSTEP / TECHNO', '-OVNI EVENT PRESENTs-', 'WWW.WEBSITE.COM', 'F WWW.YOURFACEBOOK.COM', 'T www.YOURTWITTER.com', 'YOURPLACE']", "size": [ 1275, 1875 ], "texts": [ "ANCIENT.FUTURE", "SATURDAY JUNE 30th", "ELECTRO / HOUSE / DUBSTEP / TECHNO", "-OVNI EVENT PRESENTs-", "WWW.WEBSITE.COM", "F WWW.YOURFACEBOOK.COM", "T www.YOURTWITTER.com", "YOURPLACE" ], "text_bbox": [ [ 261, 1250, 1008, 1440 ], [ 242, 1467, 1033, 1522 ], [ 256, 1546, 1026, 1577 ], [ 288, 1193, 981, 1221 ], [ 271, 1727, 1011, 1751 ], [ 737, 1602, 1026, 1630 ], [ 256, 1602, 533, 1630 ], [ 544, 1601, 728, 1638 ] ], "bbox_areas": [ 141930, 43505, 23870, 19404, 17760, 8092, 7756, 6808 ] }, { "task": "poster detection", "subtask": "", "name": "c4f6512bb6444d2ebd35a3ff33daddb4.png", "path": "poster_ocr_1024/c4f6512bb6444d2ebd35a3ff33daddb4.png", "path_original": "poster_ocr/c4f6512bb6444d2ebd35a3ff33daddb4.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['Sportswear leotard urban breathable textile make up. Runway contemporary etiquette illustration bodice sari jersey collection textile. trademark hippie fashion influence. Mainstream necessity stock embroidery retailer enhance wholesale couture motif lingerie synthetic innovation catwalk sleeveless. Quality color consumer pret-a-porter manufacture garment. Allure purchase limited clothes inexpensive. Artificial cheap comfortable. Condition hanger catwalk stitching. Artistic one-of-a-kind bargain shawl beautiful label runway popular. Couture clothing conservative. Look item color innovation Haute-couture tailor young.', 'Tailor purchase handbag imagination shawl etiquette breathable jersey conformity. Luxurious urban bold bows combination. vogue swag ribbon etiquette. Cheap pattern artistry allure jewelry. expensive vogue. Modern minimalist hanger beautiful posture production.Catwalk attractive color purchase make up elegant. Etiquette halter jeans expirement pumps apron shawl replicate look conformity independant innovation ensemble. Prediction vogue shawl hippie. Elegant imagination cut runway showcase sari make up couture industry prediction sewing. Catwalk imagination inspiration waistline outlet affection glitter. Outfit artistic glossy beautiful clothes production bold showcase cheap petticoat retailer inspiration trademark wholesale.', 'trends']", "size": [ 1200, 1200 ], "texts": [ "Sportswear leotard urban breathable textile make up. Runway contemporary etiquette illustration bodice sari jersey collection textile. trademark hippie fashion influence. Mainstream necessity stock embroidery retailer enhance wholesale couture motif lingerie synthetic innovation catwalk sleeveless. Quality color consumer pret-a-porter manufacture garment. Allure purchase limited clothes inexpensive. Artificial cheap comfortable. Condition hanger catwalk stitching. Artistic one-of-a-kind bargain shawl beautiful label runway popular. Couture clothing conservative. Look item color innovation Haute-couture tailor young.", "Tailor purchase handbag imagination shawl etiquette breathable jersey conformity. Luxurious urban bold bows combination. vogue swag ribbon etiquette. Cheap pattern artistry allure jewelry. expensive vogue. Modern minimalist hanger beautiful posture production.Catwalk attractive color purchase make up elegant. Etiquette halter jeans expirement pumps apron shawl replicate look conformity independant innovation ensemble. Prediction vogue shawl hippie. Elegant imagination cut runway showcase sari make up couture industry prediction sewing. Catwalk imagination inspiration waistline outlet affection glitter. Outfit artistic glossy beautiful clothes production bold showcase cheap petticoat retailer inspiration trademark wholesale.", "trends" ], "text_bbox": [ [ 90, 869, 648, 958 ], [ 534, 237, 1087, 322 ], [ 382, 565, 811, 635 ] ], "bbox_areas": [ 49662, 47005, 30030 ] }, { "task": "poster detection", "subtask": "", "name": "c48b6ebc33624bfcafae64e1e5408d62.png", "path": "poster_ocr_1024/c48b6ebc33624bfcafae64e1e5408d62.png", "path_original": "poster_ocr/c48b6ebc33624bfcafae64e1e5408d62.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['God A;mighty', 'God A;mighty', 'God A;mighty', 'Grace of godPraise Worship Concert Pray', 'Grace of godPraise Worship Concert Pray', 'Grace of godPraise Worship Concert Pray', 'Gospel Church', 'Gospel Church', 'Gospel Church', 'In Your Grace', 'In Your Grace', 'In Your Grace', 'In The Loving', 'In The Loving', 'In The Loving', '28 November 2016', '28 November 2016', '28 November 2016']", "size": [ 2551, 3579 ], "texts": [ "God A;mighty", "God A;mighty", "God A;mighty", "Grace of godPraise Worship Concert Pray", "Grace of godPraise Worship Concert Pray", "Grace of godPraise Worship Concert Pray", "Gospel Church", "Gospel Church", "Gospel Church", "In Your Grace", "In Your Grace", "In Your Grace", "In The Loving", "In The Loving", "In The Loving", "28 November 2016", "28 November 2016", "28 November 2016" ], "text_bbox": [ [ 305, 360, 2211, 912 ], [ 305, 360, 2211, 912 ], [ 305, 360, 2211, 912 ], [ 578, 1826, 2028, 2147 ], [ 578, 1826, 2028, 2147 ], [ 578, 1826, 2028, 2147 ], [ 753, 1545, 1871, 1827 ], [ 753, 1545, 1871, 1827 ], [ 753, 1545, 1871, 1827 ], [ 612, 2883, 2013, 3017 ], [ 612, 2883, 2013, 3017 ], [ 612, 2883, 2013, 3017 ], [ 981, 3069, 1743, 3279 ], [ 981, 3069, 1743, 3279 ], [ 981, 3069, 1743, 3279 ], [ 889, 787, 1682, 849 ], [ 889, 787, 1682, 849 ], [ 889, 787, 1682, 849 ] ], "bbox_areas": [ 1052112, 1052112, 1052112, 465450, 465450, 465450, 315276, 315276, 315276, 187734, 187734, 187734, 160020, 160020, 160020, 49166, 49166, 49166 ] }, { "task": "poster detection", "subtask": "", "name": "c40b1f89548b4492bc5b3d97a348a078.png", "path": "poster_ocr_1024/c40b1f89548b4492bc5b3d97a348a078.png", "path_original": "poster_ocr/c40b1f89548b4492bc5b3d97a348a078.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['PARTY', 'Cup', 'RED', 'flyerheroes.com presents', 'sickflyers.com / / club vibe 1235 WEST CLUB RD. ', 'with dj max']", "size": [ 1275, 1875 ], "texts": [ "PARTY", "Cup", "RED", "flyerheroes.com presents", "sickflyers.com / / club vibe 1235 WEST CLUB RD. ", "with dj max" ], "text_bbox": [ [ 243, 597, 790, 1054 ], [ 558, 152, 888, 611 ], [ 239, 224, 516, 629 ], [ 258, 148, 639, 255 ], [ 244, 1680, 1030, 1715 ], [ 730, 1573, 980, 1625 ] ], "bbox_areas": [ 249979, 151470, 112185, 40767, 27510, 13000 ] }, { "task": "poster detection", "subtask": "", "name": "c405ce3e0ce841829f71a6e8dd6296f9.png", "path": "poster_ocr_1024/c405ce3e0ce841829f71a6e8dd6296f9.png", "path_original": "poster_ocr/c405ce3e0ce841829f71a6e8dd6296f9.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['MARDI', 'GRAS', 'CARNIVAL NIGHT PARTY', 'DJ SHADE | DJ SLICK | DJ PAUL', 'ZEPPELIN PRESENTS', 'CARNIVAL MASKS MANDATORY', '20$ ENTRY FEE', 'MARCH 4', 'DOORS OPEN AT 8PM', 'PRESENTING', 'FREE FOR LADIES', 'MAIN STREET', 'TUESDAY', '555 552 255', 'THIS SPRING', 'YOUR CITY', 'WWW.YOURCLUBADRESS.COM', 'FACEBOOK.COM/YOURCLUB', 'TWITTER.COM/YOURCLUB', 'FOR RESERVATIONS', '2014']", "size": [ 2700, 3450 ], "texts": [ "MARDI", "GRAS", "CARNIVAL NIGHT PARTY", "DJ SHADE | DJ SLICK | DJ PAUL", "ZEPPELIN PRESENTS", "CARNIVAL MASKS MANDATORY", "20$ ENTRY FEE", "MARCH 4", "DOORS OPEN AT 8PM", "PRESENTING", "FREE FOR LADIES", "MAIN STREET", "TUESDAY", "555 552 255", "THIS SPRING", "YOUR CITY", "WWW.YOURCLUBADRESS.COM", "FACEBOOK.COM/YOURCLUB", "TWITTER.COM/YOURCLUB", "FOR RESERVATIONS", "2014" ], "text_bbox": [ [ 382, 328, 1377, 749 ], [ 1463, 327, 2318, 748 ], [ 385, 795, 1370, 859 ], [ 1393, 2312, 2318, 2379 ], [ 1592, 235, 2321, 298 ], [ 1482, 2693, 2318, 2738 ], [ 1781, 2436, 2318, 2506 ], [ 380, 956, 816, 1033 ], [ 1675, 2597, 2317, 2648 ], [ 1797, 2202, 2297, 2266 ], [ 1780, 2524, 2317, 2580 ], [ 1906, 2810, 2317, 2860 ], [ 380, 885, 706, 944 ], [ 1974, 3016, 2317, 3065 ], [ 1606, 160, 1968, 205 ], [ 2009, 2876, 2317, 2926 ], [ 381, 3199, 905, 3226 ], [ 1833, 3198, 2313, 3225 ], [ 1154, 3199, 1597, 3226 ], [ 1975, 2974, 2315, 3004 ], [ 379, 1042, 487, 1081 ] ], "bbox_areas": [ 418895, 359955, 63040, 61975, 45927, 37620, 37590, 33572, 32742, 32000, 30072, 20550, 19234, 16807, 16290, 15400, 14148, 12960, 11961, 10200, 4212 ] }, { "task": "poster detection", "subtask": "", "name": "c353f1aceacf49ad8dc2db78b248bc1d.png", "path": "poster_ocr_1024/c353f1aceacf49ad8dc2db78b248bc1d.png", "path_original": "poster_ocr/c353f1aceacf49ad8dc2db78b248bc1d.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['ABSINTH', 'Nights', 'NOVEMBER 7', 'Apperances by', 'LOCATED 123 SOMewHERE STREET CUTY - NEW YORK', 'renderyourmind', 'curiouspeeps', 'SEVENSTYLES ', 'SEVENSTYLES PRESENTS']", "size": [ 1275, 1875 ], "texts": [ "ABSINTH", "Nights", "NOVEMBER 7", "Apperances by", "LOCATED 123 SOMewHERE STREET CUTY - NEW YORK", "renderyourmind", "curiouspeeps", "SEVENSTYLES ", "SEVENSTYLES PRESENTS" ], "text_bbox": [ [ 395, 1001, 875, 1199 ], [ 450, 1172, 805, 1333 ], [ 448, 1470, 830, 1577 ], [ 462, 1582, 788, 1663 ], [ 365, 1764, 908, 1800 ], [ 504, 1677, 765, 1727 ], [ 845, 1677, 1055, 1727 ], [ 233, 1677, 428, 1727 ], [ 519, 948, 770, 983 ] ], "bbox_areas": [ 95040, 57155, 40874, 26406, 19548, 13050, 10500, 9750, 8785 ] }, { "task": "poster detection", "subtask": "", "name": "c2d0cdc77cf049078c005c204fa09d7a.png", "path": "poster_ocr_1024/c2d0cdc77cf049078c005c204fa09d7a.png", "path_original": "poster_ocr/c2d0cdc77cf049078c005c204fa09d7a.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['God A;mighty', 'Grace of godPraise Worship Concert Pray', 'Gospel Church', 'In Your Grace', 'In The Loving', '28 November 2016']", "size": [ 2551, 3579 ], "texts": [ "God A;mighty", "Grace of godPraise Worship Concert Pray", "Gospel Church", "In Your Grace", "In The Loving", "28 November 2016" ], "text_bbox": [ [ 305, 360, 2211, 912 ], [ 578, 1826, 2028, 2147 ], [ 753, 1545, 1871, 1827 ], [ 612, 2883, 2013, 3017 ], [ 981, 3069, 1743, 3279 ], [ 889, 787, 1682, 849 ] ], "bbox_areas": [ 1052112, 465450, 315276, 187734, 160020, 49166 ] }, { "task": "poster detection", "subtask": "", "name": "c2c100bfc5ae4e4d80d49cf66c0a1b04.png", "path": "poster_ocr_1024/c2c100bfc5ae4e4d80d49cf66c0a1b04.png", "path_original": "poster_ocr/c2c100bfc5ae4e4d80d49cf66c0a1b04.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['World Forest Day', 'Forest protection now', 'Lorem ipsum dolor sit amet, consectetur adipiscing elit sed do eiusmod tempor incididunt ut labore et dolore.', 'Lorem ipsum dolor sit amet, consectetur adipiscing elit sed do eiusmod tempor incididunt ut labore et dolore.', '21 March', 'your account', 'your account', 'your account']", "size": [ 2480, 3508 ], "texts": [ "World Forest Day", "Forest protection now", "Lorem ipsum dolor sit amet, consectetur adipiscing elit sed do eiusmod tempor incididunt ut labore et dolore.", "Lorem ipsum dolor sit amet, consectetur adipiscing elit sed do eiusmod tempor incididunt ut labore et dolore.", "21 March", "your account", "your account", "your account" ], "text_bbox": [ [ 246, 737, 450, 2641 ], [ 448, 989, 694, 2413 ], [ 257, 2941, 1386, 3168 ], [ 1507, 213, 2305, 530 ], [ 1879, 613, 2286, 693 ], [ 339, 3275, 649, 3320 ], [ 865, 3271, 1174, 3316 ], [ 1381, 3267, 1690, 3312 ] ], "bbox_areas": [ 388416, 350304, 256283, 252966, 32560, 13950, 13905, 13905 ] }, { "task": "poster detection", "subtask": "", "name": "c22c239c391c49369f00e0185eae4370.png", "path": "poster_ocr_1024/c22c239c391c49369f00e0185eae4370.png", "path_original": "poster_ocr/c22c239c391c49369f00e0185eae4370.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['SUMMER', 'MC JONES | DJ MARK | & SPECIAL GUESTS', 'BREAK', 'MAY 1TH', '1202 MAIN STREET, CITY NEW YORK, NEAR THE PLAZA', 'COCKTAILS AT THE BAR / FREE PARKING / DOORS OPEN AT 9PM', 'FACEBOOK/CLUBNAME TWITTER/CLUBNAME YOUTUBE/CLUBNAME', '@ ENVATO NIGHT CLUB', '25$', 'A GRAPHIC PRESENTS', 'ENTRY', 'WWW.WEBSITE.COM', '5566 664 443']", "size": [ 2700, 3450 ], "texts": [ "SUMMER", "MC JONES | DJ MARK | & SPECIAL GUESTS", "BREAK", "MAY 1TH", "1202 MAIN STREET, CITY NEW YORK, NEAR THE PLAZA", "COCKTAILS AT THE BAR / FREE PARKING / DOORS OPEN AT 9PM", "FACEBOOK/CLUBNAME TWITTER/CLUBNAME YOUTUBE/CLUBNAME", "@ ENVATO NIGHT CLUB", "25$", "A GRAPHIC PRESENTS", "ENTRY", "WWW.WEBSITE.COM", "5566 664 443" ], "text_bbox": [ [ 661, 1531, 2041, 1769 ], [ 576, 429, 2122, 511 ], [ 1012, 1801, 1689, 1952 ], [ 1010, 2126, 1687, 2260 ], [ 576, 2756, 2120, 2814 ], [ 576, 2699, 2126, 2745 ], [ 578, 3143, 2127, 3184 ], [ 1012, 1994, 1679, 2059 ], [ 625, 624, 894, 751 ], [ 1086, 254, 1615, 301 ], [ 623, 752, 895, 813 ], [ 1727, 2871, 2125, 2906 ], [ 1856, 2914, 2121, 2951 ] ], "bbox_areas": [ 328440, 126772, 102227, 90718, 89552, 71300, 63509, 43355, 34163, 24863, 16592, 13930, 9805 ] }, { "task": "poster detection", "subtask": "", "name": "c2275e6da6ad4f4d84858f0f0ed0586b.png", "path": "poster_ocr_1024/c2275e6da6ad4f4d84858f0f0ed0586b.png", "path_original": "poster_ocr/c2275e6da6ad4f4d84858f0f0ed0586b.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['diamondfashion', '22/03/2014', 'more info', 'your club']", "size": [ 2625, 3375 ], "texts": [ "diamondfashion", "22/03/2014", "more info", "your club" ], "text_bbox": [ [ 706, 1458, 1919, 1902 ], [ 915, 1177, 1710, 1320 ], [ 869, 2008, 1755, 2134 ], [ 898, 161, 1726, 289 ] ], "bbox_areas": [ 538572, 113685, 111636, 105984 ] }, { "task": "poster detection", "subtask": "", "name": "c1c737ceff7b47cc875149b2453a4bd0.png", "path": "poster_ocr_1024/c1c737ceff7b47cc875149b2453a4bd0.png", "path_original": "poster_ocr/c1c737ceff7b47cc875149b2453a4bd0.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['Summer', 'Sale', 'USE COUPON CODE', 'Swipe for more infe']", "size": [ 1000, 1000 ], "texts": [ "Summer", "Sale", "USE COUPON CODE", "Swipe for more infe" ], "text_bbox": [ [ 168, 84, 856, 313 ], [ 271, 231, 634, 460 ], [ 351, 908, 650, 932 ], [ 386, 941, 614, 959 ] ], "bbox_areas": [ 157552, 83127, 7176, 4104 ] }, { "task": "poster detection", "subtask": "", "name": "c17ca6e888354d9285b35ea5d9127a46.png", "path": "poster_ocr_1024/c17ca6e888354d9285b35ea5d9127a46.png", "path_original": "poster_ocr/c17ca6e888354d9285b35ea5d9127a46.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['BARBEQUE', 'PARTY', 'a day of fun in the park | family time | lots of games', '2351 main street | the community park | portland mn.', 'FACEBOOK/EVENT TWITTER/EVENT YOUTUBE/EVENT', 'free', 'LIVE', 'SATURDAY', '- The -', 'JULY 26TH', 'food', 'MUSIC']", "size": [ 2700, 3450 ], "texts": [ "BARBEQUE", "PARTY", "a day of fun in the park | family time | lots of games", "2351 main street | the community park | portland mn.", "FACEBOOK/EVENT TWITTER/EVENT YOUTUBE/EVENT", "free", "LIVE", "SATURDAY", "- The -", "JULY 26TH", "food", "MUSIC" ], "text_bbox": [ [ 309, 424, 2384, 798 ], [ 780, 725, 1924, 1007 ], [ 293, 2915, 2402, 3039 ], [ 296, 3046, 2402, 3123 ], [ 407, 3238, 2300, 3284 ], [ 372, 2166, 740, 2396 ], [ 1972, 2163, 2313, 2410 ], [ 291, 1169, 1008, 1273 ], [ 1110, 191, 1587, 347 ], [ 1701, 1170, 2387, 1274 ], [ 370, 2401, 741, 2505 ], [ 1970, 2418, 2315, 2507 ] ], "bbox_areas": [ 776050, 322608, 261516, 162162, 87078, 84640, 84227, 74568, 74412, 71344, 38584, 30705 ] }, { "task": "poster detection", "subtask": "", "name": "bfe02f8de3e44103b451d6c15a73c5d8.png", "path": "poster_ocr_1024/bfe02f8de3e44103b451d6c15a73c5d8.png", "path_original": "poster_ocr/bfe02f8de3e44103b451d6c15a73c5d8.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['dia de losmu ertos', 'musica folkloricA / altares artisticos', 'envato presents', 'entrada GRATIS + 2 MARGARITAS | PUERTAS HABREN A LAS 7:00pm', 'www.diadelosmuertos.com', 'ARTE & CULTURA']", "size": [ 1575, 2175 ], "texts": [ "dia de losmu ertos", "musica folkloricA / altares artisticos", "envato presents", "entrada GRATIS + 2 MARGARITAS | PUERTAS HABREN A LAS 7:00pm", "www.diadelosmuertos.com", "ARTE & CULTURA" ], "text_bbox": [ [ 161, 167, 1415, 913 ], [ 121, 1752, 1451, 1829 ], [ 219, 190, 744, 292 ], [ 126, 1849, 1448, 1882 ], [ 419, 1926, 1151, 1945 ], [ 571, 1703, 994, 1725 ] ], "bbox_areas": [ 935484, 102410, 53550, 43626, 13908, 9306 ] }, { "task": "poster detection", "subtask": "", "name": "bee21b8716574ed5857efc86909d0ce4.png", "path": "poster_ocr_1024/bee21b8716574ed5857efc86909d0ce4.png", "path_original": "poster_ocr/bee21b8716574ed5857efc86909d0ce4.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['UNDERGROUND', 'MONSTERNIGHT', '31st', '@6258 Amesbury StSan Diego, CA', '10:30PM', 'ANEKDAMIAN ENTERNTAINMENT PRESENTS', 'DJDAMIAN', 'DJANEK', 'WITH THE PERFORMANCES OF', 'october', '2013', 'thursday']", "size": [ 1350, 1950 ], "texts": [ "UNDERGROUND", "MONSTERNIGHT", "31st", "@6258 Amesbury StSan Diego, CA", "10:30PM", "ANEKDAMIAN ENTERNTAINMENT PRESENTS", "DJDAMIAN", "DJANEK", "WITH THE PERFORMANCES OF", "october", "2013", "thursday" ], "text_bbox": [ [ 180, 658, 1170, 772 ], [ 184, 523, 1170, 630 ], [ 573, 856, 880, 1003 ], [ 812, 1444, 1214, 1527 ], [ 833, 1563, 1205, 1641 ], [ 140, 126, 1210, 152 ], [ 149, 1580, 557, 1641 ], [ 217, 1487, 520, 1549 ], [ 143, 1445, 567, 1468 ], [ 373, 912, 594, 951 ], [ 898, 862, 947, 1000 ], [ 372, 882, 556, 911 ] ], "bbox_areas": [ 112860, 105502, 45129, 33366, 29016, 27820, 24888, 18786, 9752, 8619, 6762, 5336 ] }, { "task": "poster detection", "subtask": "", "name": "bcad2eb5626042d3a0fd0d533e5f85a2.png", "path": "poster_ocr_1024/bcad2eb5626042d3a0fd0d533e5f85a2.png", "path_original": "poster_ocr/bcad2eb5626042d3a0fd0d533e5f85a2.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['ABSINTH', 'Nights', 'NOVEMBER 7', 'Apperances by', 'LOCATED 123 SOMewHERE STREET CUTY - NEW YORK', 'renderyourmind', 'curiouspeeps', 'SEVENSTYLES ', 'SEVENSTYLES PRESENTS']", "size": [ 1275, 1875 ], "texts": [ "ABSINTH", "Nights", "NOVEMBER 7", "Apperances by", "LOCATED 123 SOMewHERE STREET CUTY - NEW YORK", "renderyourmind", "curiouspeeps", "SEVENSTYLES ", "SEVENSTYLES PRESENTS" ], "text_bbox": [ [ 395, 983, 875, 1181 ], [ 450, 1154, 805, 1315 ], [ 448, 1470, 830, 1577 ], [ 462, 1582, 788, 1663 ], [ 365, 1764, 908, 1800 ], [ 504, 1677, 765, 1727 ], [ 845, 1677, 1055, 1727 ], [ 233, 1677, 428, 1727 ], [ 519, 930, 770, 965 ] ], "bbox_areas": [ 95040, 57155, 40874, 26406, 19548, 13050, 10500, 9750, 8785 ] }, { "task": "poster detection", "subtask": "", "name": "b9c6daa0c5924c2d98dc8465591d1106.png", "path": "poster_ocr_1024/b9c6daa0c5924c2d98dc8465591d1106.png", "path_original": "poster_ocr/b9c6daa0c5924c2d98dc8465591d1106.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['ELECTRO', 'Party', 'TIESTO | midparyty |CARL COX | AIVICI', 'YOUR VENUE, YOUR STREET 1420', 'GIRLS FREE AT MIDNIGHT | 20+ ENTRY', 'FOR TICKETS & MORE INFO MAIL US AT INFO@ELECTROPARTY.COM OR VISIT OUR WEBSITE ', 'FRIDAY 21TH MAY 2016', '25$', 'GATE OPEN AT 10PM', 'eNTRY', 'FEATURING', 'WWW.ELECTROPARTY.COM', '- GIRAFFFIC PRESENTS- ']", "size": [ 2556, 3582 ], "texts": [ "ELECTRO", "Party", "TIESTO | midparyty |CARL COX | AIVICI", "YOUR VENUE, YOUR STREET 1420", "GIRLS FREE AT MIDNIGHT | 20+ ENTRY", "FOR TICKETS & MORE INFO MAIL US AT INFO@ELECTROPARTY.COM OR VISIT OUR WEBSITE ", "FRIDAY 21TH MAY 2016", "25$", "GATE OPEN AT 10PM", "eNTRY", "FEATURING", "WWW.ELECTROPARTY.COM", "- GIRAFFFIC PRESENTS- " ], "text_bbox": [ [ 238, 967, 2389, 1772 ], [ 642, 1530, 1910, 1938 ], [ 312, 2453, 2232, 2582 ], [ 540, 2704, 2009, 2784 ], [ 575, 3039, 1976, 3112 ], [ 172, 3146, 2365, 3186 ], [ 900, 1887, 1651, 1965 ], [ 452, 609, 702, 784 ], [ 990, 2877, 1558, 2947 ], [ 450, 802, 704, 899 ], [ 1071, 2316, 1480, 2375 ], [ 1005, 3376, 1540, 3420 ], [ 1028, 175, 1510, 219 ] ], "bbox_areas": [ 1731555, 517344, 247680, 117520, 102273, 87720, 58578, 43750, 39760, 24638, 24131, 23540, 21208 ] }, { "task": "poster detection", "subtask": "", "name": "b8e126fe7f064fabbf117ee3a2cbfaf9.png", "path": "poster_ocr_1024/b8e126fe7f064fabbf117ee3a2cbfaf9.png", "path_original": "poster_ocr/b8e126fe7f064fabbf117ee3a2cbfaf9.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['SATURDAY 25th OCTOBER 2014', 'THE STROKES', 'dj FOLK | Dj GIGA | DJ GRAPHICS', 'Club designers 123 St. YOUR CITY', 'THE BEST SHOW IN TOWN', 'ENTRY 10$ free admission for ladies until 10', 'NightClub', 'YOUR ENTERTAINMENT PRESENTS', 'facebook.com/YOURNAME', 'twitter.com/YOURNAME', 'www.clubname.com', 'wWw.website.com']", "size": [ 1275, 1875 ], "texts": [ "SATURDAY 25th OCTOBER 2014", "THE STROKES", "dj FOLK | Dj GIGA | DJ GRAPHICS", "Club designers 123 St. YOUR CITY", "THE BEST SHOW IN TOWN", "ENTRY 10$ free admission for ladies until 10", "NightClub", "YOUR ENTERTAINMENT PRESENTS", "facebook.com/YOURNAME", "twitter.com/YOURNAME", "www.clubname.com", "wWw.website.com" ], "text_bbox": [ [ 219, 230, 1068, 339 ], [ 325, 748, 942, 869 ], [ 228, 1358, 1040, 1438 ], [ 227, 1458, 1041, 1515 ], [ 336, 882, 909, 942 ], [ 226, 1521, 1041, 1552 ], [ 505, 1647, 768, 1730 ], [ 400, 95, 869, 122 ], [ 226, 1587, 492, 1615 ], [ 541, 1587, 788, 1615 ], [ 522, 1724, 744, 1750 ], [ 847, 1588, 1041, 1613 ] ], "bbox_areas": [ 92541, 74657, 64960, 46398, 34380, 25265, 21829, 12663, 7448, 6916, 5772, 4850 ] }, { "task": "poster detection", "subtask": "", "name": "b88fb91f318347a8a223efe74634c235.png", "path": "poster_ocr_1024/b88fb91f318347a8a223efe74634c235.png", "path_original": "poster_ocr/b88fb91f318347a8a223efe74634c235.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['vibes', 'beach', 'your fashion blog here']", "size": [ 750, 1050 ], "texts": [ "vibes", "beach", "your fashion blog here" ], "text_bbox": [ [ 2, 798, 750, 1050 ], [ 5, 641, 629, 837 ], [ 147, 84, 604, 142 ] ], "bbox_areas": [ 188496, 122304, 26506 ] }, { "task": "poster detection", "subtask": "", "name": "b76e192ad3524afebb6f8bb3a32de4c7.png", "path": "poster_ocr_1024/b76e192ad3524afebb6f8bb3a32de4c7.png", "path_original": "poster_ocr/b76e192ad3524afebb6f8bb3a32de4c7.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['diSCoAWArdS', 'Music HallLAS VEGAS', 'SATURDAY JULY 30TH', 'Celebrate The BEST PLACES TO CELEBRATE in the world', 'THE ANNUALINTERNATIONAL', '2011']", "size": [ 1264, 1772 ], "texts": [ "diSCoAWArdS", "Music HallLAS VEGAS", "SATURDAY JULY 30TH", "Celebrate The BEST PLACES TO CELEBRATE in the world", "THE ANNUALINTERNATIONAL", "2011" ], "text_bbox": [ [ 249, 444, 1015, 762 ], [ 379, 1563, 884, 1691 ], [ 288, 1313, 973, 1389 ], [ 201, 1418, 1070, 1441 ], [ 499, 342, 765, 410 ], [ 558, 789, 694, 847 ] ], "bbox_areas": [ 243588, 64640, 52060, 19987, 18088, 7888 ] }, { "task": "poster detection", "subtask": "", "name": "b41e1672026e4df0b007d7af69f1c1c2.png", "path": "poster_ocr_1024/b41e1672026e4df0b007d7af69f1c1c2.png", "path_original": "poster_ocr/b41e1672026e4df0b007d7af69f1c1c2.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['LIVE MUSIC', 'LIVE MUSIC', 'INDIE', 'INDIE', 'NIGHT', 'NIGHT', 'WHITE STEPS THE STAN', 'Friday', 'MAIN STREET YOUR CITY', 'DOORS OPEN AT 9PM', 'CROWL & BAND', 'WWW.WEBSITE.COM 555 444 666', 'Performing', '18 Feb.', 'Presents', '& SPECIAL GUEST', 'ZEPPELIN', '30$', 'Tickets', 'FACEBOOK', 'YOUTUBE', 'TWITTER', 'VIMEO']", "size": [ 2700, 3450 ], "texts": [ "LIVE MUSIC", "LIVE MUSIC", "INDIE", "INDIE", "NIGHT", "NIGHT", "WHITE STEPS THE STAN", "Friday", "MAIN STREET YOUR CITY", "DOORS OPEN AT 9PM", "CROWL & BAND", "WWW.WEBSITE.COM 555 444 666", "Performing", "18 Feb.", "Presents", "& SPECIAL GUEST", "ZEPPELIN", "30$", "Tickets", "FACEBOOK", "YOUTUBE", "TWITTER", "VIMEO" ], "text_bbox": [ [ 834, 528, 2054, 892 ], [ 834, 528, 2054, 892 ], [ 983, 941, 2012, 1294 ], [ 983, 941, 2012, 1294 ], [ 738, 1298, 1685, 1570 ], [ 738, 1298, 1685, 1570 ], [ 762, 2094, 1883, 2244 ], [ 1790, 1544, 2260, 1782 ], [ 722, 2566, 1995, 2641 ], [ 829, 2688, 1888, 2767 ], [ 840, 2008, 1509, 2119 ], [ 796, 2824, 1924, 2877 ], [ 784, 1865, 1222, 2001 ], [ 1821, 1714, 2229, 1851 ], [ 1150, 337, 1560, 448 ], [ 922, 2238, 1449, 2321 ], [ 1108, 248, 1596, 326 ], [ 500, 800, 712, 945 ], [ 471, 952, 727, 1042 ], [ 623, 3216, 909, 3257 ], [ 1541, 3217, 1796, 3258 ], [ 1123, 3215, 1348, 3256 ], [ 1988, 3216, 2156, 3257 ] ], "bbox_areas": [ 444080, 444080, 363237, 363237, 257584, 257584, 168150, 111860, 95475, 83661, 74259, 59784, 59568, 55896, 45510, 43741, 38064, 30740, 23040, 11726, 10455, 9225, 6888 ] }, { "task": "poster detection", "subtask": "", "name": "b16da988bc224500ae80ee84f48c30cd.png", "path": "poster_ocr_1024/b16da988bc224500ae80ee84f48c30cd.png", "path_original": "poster_ocr/b16da988bc224500ae80ee84f48c30cd.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['FEATURINGDIRTY LOVEBANGERZROY & DAK ', 'WORLWIDE', 'dj FLYER | Dj LIQUID | DJ GRAPHICS', 'Club designers 123 St. YOUR CITY', 'STARTMIDNIGHT 00:00', 'ENTRY 10$ free admission for ladies until 10', '-YOUR ENTERTAINMENT PRESENTS-', '25.10.14', 'F facebook.com/YOURNAME', 'T twitter.com/YOURNAME', 'W www.yourname.com']", "size": [ 1275, 1875 ], "texts": [ "FEATURINGDIRTY LOVEBANGERZROY & DAK ", "WORLWIDE", "dj FLYER | Dj LIQUID | DJ GRAPHICS", "Club designers 123 St. YOUR CITY", "STARTMIDNIGHT 00:00", "ENTRY 10$ free admission for ladies until 10", "-YOUR ENTERTAINMENT PRESENTS-", "25.10.14", "F facebook.com/YOURNAME", "T twitter.com/YOURNAME", "W www.yourname.com" ], "text_bbox": [ [ 198, 633, 787, 1121 ], [ 213, 297, 1063, 414 ], [ 205, 1360, 1068, 1438 ], [ 207, 1444, 1066, 1501 ], [ 196, 1158, 570, 1279 ], [ 207, 1509, 1063, 1540 ], [ 272, 205, 1003, 236 ], [ 201, 553, 412, 613 ], [ 203, 1557, 498, 1586 ], [ 798, 1555, 1069, 1584 ], [ 519, 1555, 774, 1583 ] ], "bbox_areas": [ 287432, 99450, 67314, 48963, 45254, 26536, 22661, 12660, 8555, 7859, 7140 ] }, { "task": "poster detection", "subtask": "", "name": "b1975abb107642f694238fe4122d7b84.png", "path": "poster_ocr_1024/b1975abb107642f694238fe4122d7b84.png", "path_original": "poster_ocr/b1975abb107642f694238fe4122d7b84.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['TECHNO', 'DJ JOKER | DJ ROBERT | DJ ZEFANYA', '2 free drinks | girls are free before 10 pm $6 admission | free parking ', 'HOUSE OF', 'TUESDAY 17TH JUNE 2013 ', 'DMBK NIGHTCLUB', 'PRESENTING']", "size": [ 1275, 1875 ], "texts": [ "TECHNO", "DJ JOKER | DJ ROBERT | DJ ZEFANYA", "2 free drinks | girls are free before 10 pm $6 admission | free parking ", "HOUSE OF", "TUESDAY 17TH JUNE 2013 ", "DMBK NIGHTCLUB", "PRESENTING" ], "text_bbox": [ [ 102, 1324, 1174, 1461 ], [ 203, 1541, 1071, 1643 ], [ 323, 1680, 937, 1796 ], [ 360, 1252, 914, 1307 ], [ 540, 281, 735, 429 ], [ 452, 165, 823, 234 ], [ 569, 107, 706, 147 ] ], "bbox_areas": [ 146864, 88536, 71224, 30470, 28860, 25599, 5480 ] }, { "task": "poster detection", "subtask": "", "name": "b142c76bbc7f4a1f8ee466b69391e8ea.png", "path": "poster_ocr_1024/b142c76bbc7f4a1f8ee466b69391e8ea.png", "path_original": "poster_ocr/b142c76bbc7f4a1f8ee466b69391e8ea.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['wednesdays', 'wednesdays', 'HOokah', 'Sickflyers.com presents', 'www,sickflyers.com', '235 WEST CLUB RD.', '@club vibe']", "size": [ 1275, 1875 ], "texts": [ "wednesdays", "wednesdays", "HOokah", "Sickflyers.com presents", "www,sickflyers.com", "235 WEST CLUB RD.", "@club vibe" ], "text_bbox": [ [ 188, 336, 1106, 901 ], [ 188, 331, 1105, 896 ], [ 153, 193, 1101, 626 ], [ 175, 228, 600, 314 ], [ 859, 1499, 1109, 1531 ], [ 157, 1554, 372, 1581 ], [ 975, 1540, 1111, 1571 ] ], "bbox_areas": [ 518670, 518105, 410484, 36550, 8000, 5805, 4216 ] }, { "task": "poster detection", "subtask": "", "name": "b0bcc9c3c1824dd4b735971020a86177.png", "path": "poster_ocr_1024/b0bcc9c3c1824dd4b735971020a86177.png", "path_original": "poster_ocr/b0bcc9c3c1824dd4b735971020a86177.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['LADIESNIGHT', 'Party', 'DJ VIAN | DJ CNUL', 'SATURDAYFEB.26', 'YOURCLUB PRESENT', '123 street main road', 'facebook.com/clubname', 'twitter.com/clubname', 'info : 123 456 7890']", "size": [ 1275, 1875 ], "texts": [ "LADIESNIGHT", "Party", "DJ VIAN | DJ CNUL", "SATURDAYFEB.26", "YOURCLUB PRESENT", "123 street main road", "facebook.com/clubname", "twitter.com/clubname", "info : 123 456 7890" ], "text_bbox": [ [ 238, 971, 1031, 1307 ], [ 373, 1272, 902, 1486 ], [ 305, 1527, 976, 1604 ], [ 404, 1626, 871, 1666 ], [ 345, 97, 930, 125 ], [ 402, 1676, 870, 1705 ], [ 189, 1770, 599, 1789 ], [ 709, 1770, 1086, 1789 ], [ 472, 1712, 803, 1731 ] ], "bbox_areas": [ 266448, 113206, 51667, 18680, 16380, 13572, 7790, 7163, 6289 ] }, { "task": "poster detection", "subtask": "", "name": "abf24dcd42f84885a8ef8df3fef0d225.png", "path": "poster_ocr_1024/abf24dcd42f84885a8ef8df3fef0d225.png", "path_original": "poster_ocr/abf24dcd42f84885a8ef8df3fef0d225.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['Indie', 'Weekend', 'SATURDAY JUNE 21', 'ENTRY FEE 30$ | DRINKS SPECIAL 3$ SHOTS | DOORS OPEN AT 9PM', 'STEVENSARMADA', '8442 MAIN STREET, YOUR CITY CA. NEAR THE PLAZA', 'DONNIESTARKS', 'BRAXIS&BAND', 'The', 'FEATURING ALL HOT BANDS AND SPECIAL GUEST', 'LIVE MUSIC SHOW AT CLUB ZEPPELIN', 'WWW.WEBSITE.COM', 'FACEBOOK/CLUBNAME', 'TWITTER/CLUBNAME', 'FROM 02 TO 04', 'FROM 12 TO 02', 'FROM 9 TO 12']", "size": [ 2700, 3450 ], "texts": [ "Indie", "Weekend", "SATURDAY JUNE 21", "ENTRY FEE 30$ | DRINKS SPECIAL 3$ SHOTS | DOORS OPEN AT 9PM", "STEVENSARMADA", "8442 MAIN STREET, YOUR CITY CA. NEAR THE PLAZA", "DONNIESTARKS", "BRAXIS&BAND", "The", "FEATURING ALL HOT BANDS AND SPECIAL GUEST", "LIVE MUSIC SHOW AT CLUB ZEPPELIN", "WWW.WEBSITE.COM", "FACEBOOK/CLUBNAME", "TWITTER/CLUBNAME", "FROM 02 TO 04", "FROM 12 TO 02", "FROM 9 TO 12" ], "text_bbox": [ [ 313, 809, 2068, 1563 ], [ 859, 1361, 2379, 1890 ], [ 838, 475, 1863, 647 ], [ 546, 2603, 2154, 2708 ], [ 1167, 2225, 1530, 2480 ], [ 796, 2804, 1902, 2880 ], [ 1928, 2225, 2250, 2481 ], [ 448, 2219, 747, 2476 ], [ 1137, 800, 1471, 1005 ], [ 870, 1963, 1834, 2026 ], [ 933, 279, 1765, 350 ], [ 1048, 2891, 1656, 2984 ], [ 1588, 3169, 1924, 3212 ], [ 2035, 3169, 2347, 3212 ], [ 2006, 2157, 2206, 2199 ], [ 1252, 2157, 1445, 2199 ], [ 511, 2151, 685, 2193 ] ], "bbox_areas": [ 1323270, 804080, 176300, 168840, 92565, 84056, 82432, 76843, 68470, 60732, 59072, 56544, 14448, 13416, 8400, 8106, 7308 ] }, { "task": "poster detection", "subtask": "", "name": "ab25f11b07b742caaeccfbbb0a588451.png", "path": "poster_ocr_1024/ab25f11b07b742caaeccfbbb0a588451.png", "path_original": "poster_ocr/ab25f11b07b742caaeccfbbb0a588451.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['Photographer', 'Sports Tournament', 'PASS', 'Proin vehicula diam at turpis pretium at interdum an.', 'SAT.March29. 2012', 'Mona BaySports Club', 'Farquarson', 'Phasellus ornare portti']", "size": [ 938, 1444 ], "texts": [ "Photographer", "Sports Tournament", "PASS", "Proin vehicula diam at turpis pretium at interdum an.", "SAT.March29. 2012", "Mona BaySports Club", "Farquarson", "Phasellus ornare portti" ], "text_bbox": [ [ 76, 852, 865, 962 ], [ 159, 166, 773, 289 ], [ 113, 1008, 205, 1317 ], [ 289, 1351, 853, 1380 ], [ 295, 1187, 447, 1292 ], [ 296, 1026, 523, 1093 ], [ 384, 697, 562, 744 ], [ 836, 1034, 866, 1291 ] ], "bbox_areas": [ 86790, 75522, 28428, 16356, 15960, 15209, 8366, 7710 ] }, { "task": "poster detection", "subtask": "", "name": "aa80a5ccf7294c838e438255b44c3c7d.png", "path": "poster_ocr_1024/aa80a5ccf7294c838e438255b44c3c7d.png", "path_original": "poster_ocr/aa80a5ccf7294c838e438255b44c3c7d.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['Travel Tips', 'What you needWhat you have to prepareLow budget destination.', 'ON THEBLOG TODAY']", "size": [ 1000, 1000 ], "texts": [ "Travel Tips", "What you needWhat you have to prepareLow budget destination.", "ON THEBLOG TODAY" ], "text_bbox": [ [ 53, 53, 753, 290 ], [ 53, 826, 403, 932 ], [ 54, 693, 282, 781 ] ], "bbox_areas": [ 165900, 37100, 20064 ] }, { "task": "poster detection", "subtask": "", "name": "a955cfd3e54c4770bf16c969f3b39496.png", "path": "poster_ocr_1024/a955cfd3e54c4770bf16c969f3b39496.png", "path_original": "poster_ocr/a955cfd3e54c4770bf16c969f3b39496.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['AngelMascara', 'buy oneget onefree', 'Lorem ipsum dolor sit amet, consectetur adipiscing elit. Cras efficitur nec leo ut aliquet. Proin ut enim id libero fringilla vestibulum luctus pharetra diam. ', 'big sale!', 'www.shopmascara.com', 'Online Shopping']", "size": [ 1753, 2480 ], "texts": [ "AngelMascara", "buy oneget onefree", "Lorem ipsum dolor sit amet, consectetur adipiscing elit. Cras efficitur nec leo ut aliquet. Proin ut enim id libero fringilla vestibulum luctus pharetra diam. ", "big sale!", "www.shopmascara.com", "Online Shopping" ], "text_bbox": [ [ 93, 211, 1661, 505 ], [ 1008, 1153, 1602, 1434 ], [ 104, 716, 800, 941 ], [ 114, 1005, 824, 1081 ], [ 603, 2328, 1150, 2379 ], [ 100, 626, 574, 663 ] ], "bbox_areas": [ 460992, 166914, 156600, 53960, 27897, 17538 ] }, { "task": "poster detection", "subtask": "", "name": "a8ebc0facd794f39b9e14ad0025353ff.png", "path": "poster_ocr_1024/a8ebc0facd794f39b9e14ad0025353ff.png", "path_original": "poster_ocr/a8ebc0facd794f39b9e14ad0025353ff.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['spring', 'break!', 'Caribbean', 'MAIN ROOM: LOUIS TWELVE& ON THE PATIO: DJ STARK', 'TUESDAY, FebRuAry 17', 'your entertainment presents', 'DOORS OPEN AT 9PM / ENTRY $39 / FREE DRINKS ALL NIGHT ', 'RSVP: 123 4567 890 //www.Yourclub.com', 'LADIES', 'SPECIAL PERFORMANCE BY:', 'NO COVER']", "size": [ 1275, 1875 ], "texts": [ "spring", "break!", "Caribbean", "MAIN ROOM: LOUIS TWELVE& ON THE PATIO: DJ STARK", "TUESDAY, FebRuAry 17", "your entertainment presents", "DOORS OPEN AT 9PM / ENTRY $39 / FREE DRINKS ALL NIGHT ", "RSVP: 123 4567 890 //www.Yourclub.com", "LADIES", "SPECIAL PERFORMANCE BY:", "NO COVER" ], "text_bbox": [ [ 170, 997, 1105, 1192 ], [ 291, 1207, 988, 1366 ], [ 309, 1634, 966, 1719 ], [ 356, 1449, 918, 1534 ], [ 394, 833, 856, 901 ], [ 209, 938, 1047, 975 ], [ 193, 1571, 1077, 1602 ], [ 292, 1746, 982, 1769 ], [ 982, 235, 1112, 304 ], [ 437, 1407, 835, 1429 ], [ 971, 288, 1106, 349 ] ], "bbox_areas": [ 182325, 110823, 55845, 47770, 31416, 31006, 27404, 15870, 8970, 8756, 8235 ] }, { "task": "poster detection", "subtask": "", "name": "a87c588e8909443e9e16f1222bd32f7f.png", "path": "poster_ocr_1024/a87c588e8909443e9e16f1222bd32f7f.png", "path_original": "poster_ocr/a87c588e8909443e9e16f1222bd32f7f.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['DOPE', 'GANG', 'JOE CRACK . DJ KILL . RON BRONZ', '26/06/2016', 'YOURCLUB', 'RAP / TRAP / RNB/ DIRTY', 'HYPE STYLE PRESENTS']", "size": [ 1275, 1875 ], "texts": [ "DOPE", "GANG", "JOE CRACK . DJ KILL . RON BRONZ", "26/06/2016", "YOURCLUB", "RAP / TRAP / RNB/ DIRTY", "HYPE STYLE PRESENTS" ], "text_bbox": [ [ 284, 715, 992, 946 ], [ 282, 952, 989, 1147 ], [ 318, 1506, 950, 1556 ], [ 447, 243, 833, 308 ], [ 445, 1571, 832, 1630 ], [ 317, 1170, 960, 1202 ], [ 320, 662, 959, 694 ] ], "bbox_areas": [ 163548, 137865, 31600, 25090, 22833, 20576, 20448 ] }, { "task": "poster detection", "subtask": "", "name": "a5d3ddb25ad54f8b8f274df320db19af.png", "path": "poster_ocr_1024/a5d3ddb25ad54f8b8f274df320db19af.png", "path_original": "poster_ocr/a5d3ddb25ad54f8b8f274df320db19af.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['MusicSpecial', 'BUSINESSINSPIRN DEICATINS', '$10', 'Call us :+098 123 4556 234', 'DJ Low-J & Yolk-X']", "size": [ 1275, 1875 ], "texts": [ "MusicSpecial", "BUSINESSINSPIRN DEICATINS", "$10", "Call us :+098 123 4556 234", "DJ Low-J & Yolk-X" ], "text_bbox": [ [ 272, 705, 981, 1025 ], [ 305, 1017, 959, 1170 ], [ 439, 1196, 775, 1352 ], [ 126, 145, 413, 210 ], [ 414, 618, 833, 656 ] ], "bbox_areas": [ 226880, 100062, 52416, 18655, 15922 ] }, { "task": "poster detection", "subtask": "", "name": "a5cc21ddf0a54e5d97f6f269247aaad0.png", "path": "poster_ocr_1024/a5cc21ddf0a54e5d97f6f269247aaad0.png", "path_original": "poster_ocr/a5cc21ddf0a54e5d97f6f269247aaad0.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['Xmas', 'Candy', 'San Graphics Presents', 'YOUR CLUB', 'wed25dec', 'DANDENONG RD. NOBLE PARK // MELBOURNE // 97 000 459', 'DJ SANTA // DJ LOLLYPOP // DJ CANDY', 'DOORS OPEN 9PM']", "size": [ 1825, 2551 ], "texts": [ "Xmas", "Candy", "San Graphics Presents", "YOUR CLUB", "wed25dec", "DANDENONG RD. NOBLE PARK // MELBOURNE // 97 000 459", "DJ SANTA // DJ LOLLYPOP // DJ CANDY", "DOORS OPEN 9PM" ], "text_bbox": [ [ 445, 1595, 1247, 1982 ], [ 483, 1427, 1380, 1754 ], [ 527, 103, 1346, 196 ], [ 646, 2258, 1180, 2376 ], [ 702, 2023, 1130, 2142 ], [ 405, 2413, 1419, 2451 ], [ 539, 2179, 1285, 2228 ], [ 769, 1905, 1056, 1942 ] ], "bbox_areas": [ 310374, 293319, 76167, 63012, 50932, 38532, 36554, 10619 ] }, { "task": "poster detection", "subtask": "", "name": "a58adc56b1da4127845593908c7cfb76.png", "path": "poster_ocr_1024/a58adc56b1da4127845593908c7cfb76.png", "path_original": "poster_ocr/a58adc56b1da4127845593908c7cfb76.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['Come celebrate', 'Celebrating our new launch of our athletic wear line. Come party with us at our pop up shop in LA', 'Launch party', 'satmay 15', '9 am -5 pm']", "size": [ 750, 1050 ], "texts": [ "Come celebrate", "Celebrating our new launch of our athletic wear line. Come party with us at our pop up shop in LA", "Launch party", "satmay 15", "9 am -5 pm" ], "text_bbox": [ [ 90, 35, 661, 133 ], [ 198, 259, 555, 339 ], [ 214, 188, 541, 212 ], [ 66, 180, 160, 225 ], [ 595, 180, 683, 225 ] ], "bbox_areas": [ 55958, 28560, 7848, 4230, 3960 ] }, { "task": "poster detection", "subtask": "", "name": "a562eee227f44176ae38978dbb0cf4a6.png", "path": "poster_ocr_1024/a562eee227f44176ae38978dbb0cf4a6.png", "path_original": "poster_ocr/a562eee227f44176ae38978dbb0cf4a6.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['hour', 'happy', 'Every friday', 'mixeddrinks', 'Craftbeers', 'place presents', 'your awesome', 'from 6 to 9 pm', '123 your street 123-456-7890', '6', '5', 'from', 'from']", "size": [ 1275, 1875 ], "texts": [ "hour", "happy", "Every friday", "mixeddrinks", "Craftbeers", "place presents", "your awesome", "from 6 to 9 pm", "123 your street 123-456-7890", "6", "5", "from", "from" ], "text_bbox": [ [ 187, 647, 1071, 1163 ], [ 206, 375, 1053, 709 ], [ 256, 1093, 1009, 1291 ], [ 828, 1455, 1121, 1704 ], [ 177, 1440, 441, 1696 ], [ 811, 87, 1166, 228 ], [ 113, 109, 482, 243 ], [ 468, 1256, 806, 1363 ], [ 300, 1746, 971, 1797 ], [ 461, 1427, 582, 1599 ], [ 695, 1428, 812, 1598 ], [ 696, 1573, 801, 1642 ], [ 486, 1579, 588, 1642 ] ], "bbox_areas": [ 456144, 282898, 149094, 72957, 67584, 50055, 49446, 36166, 34221, 20812, 19890, 7245, 6426 ] }, { "task": "poster detection", "subtask": "", "name": "a4a453eac5f74c4bbb03fb5526e4a8f9.png", "path": "poster_ocr_1024/a4a453eac5f74c4bbb03fb5526e4a8f9.png", "path_original": "poster_ocr/a4a453eac5f74c4bbb03fb5526e4a8f9.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['WorldForest Day', 'Lorem ipsum dolor sit amet, consectetur adipiscing elit sed do eiusmod tempor incididunt ut labore et dolore.', 'Forest protection now', '21 March', 'www.yoursite.com', 'your account', 'your account', 'your account']", "size": [ 1753, 2480 ], "texts": [ "WorldForest Day", "Lorem ipsum dolor sit amet, consectetur adipiscing elit sed do eiusmod tempor incididunt ut labore et dolore.", "Forest protection now", "21 March", "www.yoursite.com", "your account", "your account", "your account" ], "text_bbox": [ [ 788, 936, 1661, 1272 ], [ 906, 1424, 1657, 1680 ], [ 742, 1187, 1649, 1362 ], [ 1287, 799, 1650, 869 ], [ 1183, 1782, 1591, 1807 ], [ 830, 2334, 1018, 2362 ], [ 1147, 2331, 1334, 2359 ], [ 1459, 2329, 1646, 2357 ] ], "bbox_areas": [ 293328, 192256, 158725, 25410, 10200, 5264, 5236, 5236 ] }, { "task": "poster detection", "subtask": "", "name": "a4318a4cb8664469856a98c50a0cd467.png", "path": "poster_ocr_1024/a4318a4cb8664469856a98c50a0cd467.png", "path_original": "poster_ocr/a4318a4cb8664469856a98c50a0cd467.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['For You', 'We,TongRo Image Stock, since commence with producing digital Image slide transparencybusiness in 1992, have been building outstanding success in distributingvarious kind of collections from overseas countries to Korea and insupplying our own collections to many countries througharound 45 channel partners', 'Invitation', 'Saturday.12.March', 'Tongro Convention Center', '5:00 pm', 'We are very happy to invite you']", "size": [ 3543, 4961 ], "texts": [ "For You", "We,TongRo Image Stock, since commence with producing digital Image slide transparencybusiness in 1992, have been building outstanding success in distributingvarious kind of collections from overseas countries to Korea and insupplying our own collections to many countries througharound 45 channel partners", "Invitation", "Saturday.12.March", "Tongro Convention Center", "5:00 pm", "We are very happy to invite you" ], "text_bbox": [ [ 1031, 1865, 2583, 2124 ], [ 1203, 3538, 2455, 3785 ], [ 1018, 1440, 2585, 1631 ], [ 1160, 2910, 2443, 3046 ], [ 1156, 3362, 2447, 3461 ], [ 1495, 3129, 2139, 3285 ], [ 1035, 1738, 2561, 1799 ] ], "bbox_areas": [ 401968, 309244, 299297, 174488, 127809, 100464, 93086 ] }, { "task": "poster detection", "subtask": "", "name": "a1b1d4232e65403b9ce5412d75f6078f.png", "path": "poster_ocr_1024/a1b1d4232e65403b9ce5412d75f6078f.png", "path_original": "poster_ocr/a1b1d4232e65403b9ce5412d75f6078f.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['take an extra40% off sale items', 'use code happysale* at checkoutto receive your discount', 'start shopping', 'spring sale styles']", "size": [ 750, 1050 ], "texts": [ "take an extra40% off sale items", "use code happysale* at checkoutto receive your discount", "start shopping", "spring sale styles" ], "text_bbox": [ [ 97, 409, 660, 516 ], [ 126, 561, 627, 607 ], [ 243, 668, 507, 686 ], [ 234, 348, 519, 364 ] ], "bbox_areas": [ 60241, 23046, 4752, 4560 ] }, { "task": "poster detection", "subtask": "", "name": "a05f46f8ce414c928f667f052bf86a96.png", "path": "poster_ocr_1024/a05f46f8ce414c928f667f052bf86a96.png", "path_original": "poster_ocr/a05f46f8ce414c928f667f052bf86a96.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['SHAPES', 'The', '19.04.2015', '-FAT MINIMAL MUSIC PRESENTS-', 'DEEJAY MINIMAL . DEEJAY SOUND', 'ELECTRO / HOUSE / DUBSTEP', 'YOURCLUB', 'www.yourwebsite.com']", "size": [ 1275, 1875 ], "texts": [ "SHAPES", "The", "19.04.2015", "-FAT MINIMAL MUSIC PRESENTS-", "DEEJAY MINIMAL . DEEJAY SOUND", "ELECTRO / HOUSE / DUBSTEP", "YOURCLUB", "www.yourwebsite.com" ], "text_bbox": [ [ 341, 372, 951, 488 ], [ 553, 259, 743, 350 ], [ 495, 1361, 782, 1421 ], [ 413, 194, 858, 224 ], [ 410, 1470, 861, 1499 ], [ 425, 1508, 851, 1531 ], [ 539, 1575, 738, 1621 ], [ 511, 1629, 760, 1653 ] ], "bbox_areas": [ 70760, 17290, 17220, 13350, 13079, 9798, 9154, 5976 ] }, { "task": "poster detection", "subtask": "", "name": "9fa945a5a60347ce9c8c7e1113e22a18.png", "path": "poster_ocr_1024/9fa945a5a60347ce9c8c7e1113e22a18.png", "path_original": "poster_ocr/9fa945a5a60347ce9c8c7e1113e22a18.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['protect ourForest ', 'World forest day', 'Lorem ipsum dolor sit amet, consectetur adipiscing elit sed do eiusmod tempor incididunt ut labore et dolore.', '21 March', 'www.yoursite.com', 'your account', 'your account', 'your account']", "size": [ 2480, 3508 ], "texts": [ "protect ourForest ", "World forest day", "Lorem ipsum dolor sit amet, consectetur adipiscing elit sed do eiusmod tempor incididunt ut labore et dolore.", "21 March", "www.yoursite.com", "your account", "your account", "your account" ], "text_bbox": [ [ 328, 493, 1641, 937 ], [ 341, 807, 1411, 1214 ], [ 1063, 2900, 2190, 3127 ], [ 325, 1304, 951, 1422 ], [ 380, 1597, 1104, 1646 ], [ 1107, 3304, 1509, 3363 ], [ 424, 3309, 826, 3368 ], [ 1778, 3298, 2180, 3357 ] ], "bbox_areas": [ 582972, 435490, 255829, 73868, 35476, 23718, 23718, 23718 ] }, { "task": "poster detection", "subtask": "", "name": "9f77138c65ec442d8ea20b5b70e041a2.png", "path": "poster_ocr_1024/9f77138c65ec442d8ea20b5b70e041a2.png", "path_original": "poster_ocr/9f77138c65ec442d8ea20b5b70e041a2.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['MARDI ', 'Gras', 'CARNIVAL', 'MC BUBA | DJ ROMEO | DJ JULIO | DJ MARK', 'ZEPPELIN', 'PARTY', 'DOORS OPEN AT 10 PM | 10$ ENTRY | COSTUME MANDATORY', 'Presents', 'MAIN STREET | YOUR CITY | 222 333 444', 'WWW.YOURCLUBADRESS.COM', 'FACEBOOK.COM/YOURCLUB', 'TWITTER.COM/YOURCLUB']", "size": [ 2700, 3450 ], "texts": [ "MARDI ", "Gras", "CARNIVAL", "MC BUBA | DJ ROMEO | DJ JULIO | DJ MARK", "ZEPPELIN", "PARTY", "DOORS OPEN AT 10 PM | 10$ ENTRY | COSTUME MANDATORY", "Presents", "MAIN STREET | YOUR CITY | 222 333 444", "WWW.YOURCLUBADRESS.COM", "FACEBOOK.COM/YOURCLUB", "TWITTER.COM/YOURCLUB" ], "text_bbox": [ [ 491, 1479, 2202, 2261 ], [ 1168, 1892, 2439, 2596 ], [ 557, 2428, 1259, 2625 ], [ 560, 2733, 2137, 2806 ], [ 1640, 245, 2194, 437 ], [ 1682, 2429, 2132, 2623 ], [ 564, 2829, 2135, 2879 ], [ 1817, 363, 2284, 523 ], [ 740, 2957, 1913, 3014 ], [ 495, 3269, 929, 3311 ], [ 1802, 3269, 2199, 3314 ], [ 1196, 3269, 1570, 3314 ] ], "bbox_areas": [ 1338002, 894784, 138294, 115121, 106368, 87300, 78550, 74720, 66861, 18228, 17865, 16830 ] }, { "task": "poster detection", "subtask": "", "name": "9eb23f388bdc4323b68dc2b0bd987df4.png", "path": "poster_ocr_1024/9eb23f388bdc4323b68dc2b0bd987df4.png", "path_original": "poster_ocr/9eb23f388bdc4323b68dc2b0bd987df4.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['DARKNIGHT', 'dj SKULL | Dj MONEY | DJ STRIKE', '9 PM || ENTRY 20$ || DRESS CODE : ALL IN BLACK', '-OVNI EVENT PRESENTs-', 'NightClub', 'FRIDAY 15th JULY', 'f facebook.com/YOURNAME', 'T twitter.com/YOURNAME']", "size": [ 1275, 1875 ], "texts": [ "DARKNIGHT", "dj SKULL | Dj MONEY | DJ STRIKE", "9 PM || ENTRY 20$ || DRESS CODE : ALL IN BLACK", "-OVNI EVENT PRESENTs-", "NightClub", "FRIDAY 15th JULY", "f facebook.com/YOURNAME", "T twitter.com/YOURNAME" ], "text_bbox": [ [ 148, 306, 1176, 648 ], [ 188, 1512, 1085, 1571 ], [ 185, 1594, 1092, 1633 ], [ 218, 188, 1071, 218 ], [ 510, 1651, 778, 1712 ], [ 473, 121, 815, 166 ], [ 203, 1651, 506, 1683 ], [ 789, 1651, 1069, 1683 ] ], "bbox_areas": [ 351576, 52923, 35373, 25590, 16348, 15390, 9696, 8960 ] }, { "task": "poster detection", "subtask": "", "name": "9d98cbe07a3741c6b07bad9825f06872.png", "path": "poster_ocr_1024/9d98cbe07a3741c6b07bad9825f06872.png", "path_original": "poster_ocr/9d98cbe07a3741c6b07bad9825f06872.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['Candy', 'Party', 'dj FLYER | Dj LOVER | DJ GRAPHICS', 'Club designers 123 St. YOUR CITY', 'SATURDAY 25th OCTOBER 2014', 'ENTRY 10$ free admission for ladies until 10', 'NightClub', 'YOUR ENTERTAINMENT PRESENTS', 'facebook.com/YOURNAME', 'twitter.com/YOURNAME', 'www.clubname.com']", "size": [ 1311, 1819 ], "texts": [ "Candy", "Party", "dj FLYER | Dj LOVER | DJ GRAPHICS", "Club designers 123 St. YOUR CITY", "SATURDAY 25th OCTOBER 2014", "ENTRY 10$ free admission for ladies until 10", "NightClub", "YOUR ENTERTAINMENT PRESENTS", "facebook.com/YOURNAME", "twitter.com/YOURNAME", "www.clubname.com" ], "text_bbox": [ [ 283, 650, 1061, 1187 ], [ 317, 1003, 867, 1377 ], [ 242, 1372, 1070, 1456 ], [ 236, 1462, 1077, 1522 ], [ 305, 127, 998, 199 ], [ 236, 1532, 1074, 1564 ], [ 529, 1606, 792, 1689 ], [ 303, 91, 1001, 119 ], [ 215, 1644, 490, 1674 ], [ 873, 1644, 1128, 1674 ], [ 547, 1685, 769, 1711 ] ], "bbox_areas": [ 417786, 205700, 69552, 50460, 49896, 26816, 21829, 19544, 8250, 7650, 5772 ] }, { "task": "poster detection", "subtask": "", "name": "9b2ca9c475e54c48a8b39c6ba1196bfd.png", "path": "poster_ocr_1024/9b2ca9c475e54c48a8b39c6ba1196bfd.png", "path_original": "poster_ocr/9b2ca9c475e54c48a8b39c6ba1196bfd.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['Glow', 'Glow', 'Party', 'Party', 'FRIDAY, MARCH 21', 'FRIDAY, MARCH 21', '20$ ENTRY | FREE FOR GIRLS | DOORS OPEN AT 9PM ', 'MAIN STREET | YOUR CITY | NEAR THE PLAZA', 'FOR MORE INFO: WWW.WEBSITE.COM | 444 555 222', 'ZEPPELIN PRESENTS', 'DJ SHADE', 'DJ SLICK', 'DJ PAUL', 'FACEBOOK', 'YOUTUBE', 'TWITTER', 'VIMEO']", "size": [ 2700, 3450 ], "texts": [ "Glow", "Glow", "Party", "Party", "FRIDAY, MARCH 21", "FRIDAY, MARCH 21", "20$ ENTRY | FREE FOR GIRLS | DOORS OPEN AT 9PM ", "MAIN STREET | YOUR CITY | NEAR THE PLAZA", "FOR MORE INFO: WWW.WEBSITE.COM | 444 555 222", "ZEPPELIN PRESENTS", "DJ SHADE", "DJ SLICK", "DJ PAUL", "FACEBOOK", "YOUTUBE", "TWITTER", "VIMEO" ], "text_bbox": [ [ 954, 1436, 1996, 1904 ], [ 954, 1436, 1996, 1904 ], [ 671, 1768, 1893, 2146 ], [ 671, 1768, 1893, 2146 ], [ 533, 2429, 2171, 2599 ], [ 533, 2429, 2171, 2599 ], [ 336, 2742, 2358, 2818 ], [ 389, 2846, 2303, 2918 ], [ 548, 3013, 2143, 3065 ], [ 861, 193, 1836, 271 ], [ 1079, 434, 1613, 525 ], [ 292, 434, 765, 525 ], [ 1943, 434, 2411, 525 ], [ 623, 3216, 909, 3257 ], [ 1541, 3217, 1796, 3258 ], [ 1123, 3215, 1348, 3256 ], [ 1988, 3216, 2156, 3257 ] ], "bbox_areas": [ 487656, 487656, 461916, 461916, 278460, 278460, 153672, 137808, 82940, 76050, 48594, 43043, 42588, 11726, 10455, 9225, 6888 ] }, { "task": "poster detection", "subtask": "", "name": "9a261dbdb5c9455197f762487995f9e1.png", "path": "poster_ocr_1024/9a261dbdb5c9455197f762487995f9e1.png", "path_original": "poster_ocr/9a261dbdb5c9455197f762487995f9e1.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['LiveSupport band namemusic band name', 'MoustacheParty', 'Got Mo?Get Free Enter', 'Your Place Name. Street 12/Dwww.movemberparty.com ', '30.Nov.15']", "size": [ 2625, 3375 ], "texts": [ "LiveSupport band namemusic band name", "MoustacheParty", "Got Mo?Get Free Enter", "Your Place Name. Street 12/Dwww.movemberparty.com ", "30.Nov.15" ], "text_bbox": [ [ 314, 2287, 2280, 2828 ], [ 442, 1124, 2164, 1722 ], [ 426, 441, 2206, 955 ], [ 648, 2983, 1953, 3205 ], [ 967, 207, 1636, 318 ] ], "bbox_areas": [ 1063606, 1029756, 914920, 289710, 74259 ] }, { "task": "poster detection", "subtask": "", "name": "97c06d403a6c4d37a4ef50f37ba30a9b.png", "path": "poster_ocr_1024/97c06d403a6c4d37a4ef50f37ba30a9b.png", "path_original": "poster_ocr/97c06d403a6c4d37a4ef50f37ba30a9b.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['THESUMMERTRENCH', 'Our most popular trench coat for summer is 50% off when you use code 50TRENCH at checkoutGET IT NOW']", "size": [ 750, 1050 ], "texts": [ "THESUMMERTRENCH", "Our most popular trench coat for summer is 50% off when you use code 50TRENCH at checkoutGET IT NOW" ], "text_bbox": [ [ 137, 201, 588, 525 ], [ 128, 820, 627, 1013 ] ], "bbox_areas": [ 146124, 96307 ] }, { "task": "poster detection", "subtask": "", "name": "977570f6fdb1440481877e0183fd0836.png", "path": "poster_ocr_1024/977570f6fdb1440481877e0183fd0836.png", "path_original": "poster_ocr/977570f6fdb1440481877e0183fd0836.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['Body', 'DETOX PLAN']", "size": [ 1080, 1920 ], "texts": [ "Body", "DETOX PLAN" ], "text_bbox": [ [ 304, 735, 770, 1177 ], [ 169, 1038, 916, 1106 ] ], "bbox_areas": [ 205972, 50796 ] }, { "task": "poster detection", "subtask": "", "name": "96d6a3471b9b42539bdda710fd2488c1.png", "path": "poster_ocr_1024/96d6a3471b9b42539bdda710fd2488c1.png", "path_original": "poster_ocr/96d6a3471b9b42539bdda710fd2488c1.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['TITANS', 'SATURDAY JUNE 30th', 'ELECTRO / HOUSE / DUBSTEP / TECHNO', '-OVNI EVENT PRESENTs-', 'WWW.WEBSITE.COM', 'F WWW.YOURFACEBOOK.COM', 'T www.YOURTWITTER.com', 'YOURPLACE']", "size": [ 1275, 1875 ], "texts": [ "TITANS", "SATURDAY JUNE 30th", "ELECTRO / HOUSE / DUBSTEP / TECHNO", "-OVNI EVENT PRESENTs-", "WWW.WEBSITE.COM", "F WWW.YOURFACEBOOK.COM", "T www.YOURTWITTER.com", "YOURPLACE" ], "text_bbox": [ [ 237, 1307, 1034, 1456 ], [ 245, 1491, 1036, 1546 ], [ 259, 1570, 1029, 1601 ], [ 288, 1247, 981, 1275 ], [ 274, 1751, 1014, 1775 ], [ 740, 1626, 1029, 1654 ], [ 259, 1626, 536, 1654 ], [ 547, 1625, 731, 1662 ] ], "bbox_areas": [ 118753, 43505, 23870, 19404, 17760, 8092, 7756, 6808 ] }, { "task": "poster detection", "subtask": "", "name": "93f6a3d3ec0448b4a324ab7338e34132.png", "path": "poster_ocr_1024/93f6a3d3ec0448b4a324ab7338e34132.png", "path_original": "poster_ocr/93f6a3d3ec0448b4a324ab7338e34132.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['electrofuture', 'electro sound Partygarfi don | JOHN DOE | STEPHEN JACK ', 'Lorem Ipsum is simply dummy text of the printing and typesetting industry. galley of type and scrambled it to make special.Lorem Ipsum is simply dummy text of the prin', 'special guestdj daviddj garfild', 'JUNUARY15-12-2012-9PM', 'OUR BEST MUSIC PARTNERS IN THIS PARTY 2015 DECEMBER', 'exclusive PRESENTS', '25.04.15', 'SAT 15-12-2015', 'TICKET ARE AVAILABLE IN OUR STOCK', 'performing']", "size": [ 1350, 1950 ], "texts": [ "electrofuture", "electro sound Partygarfi don | JOHN DOE | STEPHEN JACK ", "Lorem Ipsum is simply dummy text of the printing and typesetting industry. galley of type and scrambled it to make special.Lorem Ipsum is simply dummy text of the prin", "special guestdj daviddj garfild", "JUNUARY15-12-2012-9PM", "OUR BEST MUSIC PARTNERS IN THIS PARTY 2015 DECEMBER", "exclusive PRESENTS", "25.04.15", "SAT 15-12-2015", "TICKET ARE AVAILABLE IN OUR STOCK", "performing" ], "text_bbox": [ [ 395, 579, 959, 762 ], [ 225, 1583, 1122, 1684 ], [ 186, 1700, 1165, 1755 ], [ 486, 892, 862, 1012 ], [ 106, 80, 444, 161 ], [ 166, 1839, 1175, 1865 ], [ 434, 1776, 917, 1808 ], [ 551, 799, 799, 846 ], [ 955, 77, 1274, 103 ], [ 859, 118, 1275, 135 ], [ 509, 1524, 840, 1541 ] ], "bbox_areas": [ 103212, 90597, 53845, 45120, 27378, 26234, 15456, 11656, 8294, 7072, 5627 ] }, { "task": "poster detection", "subtask": "", "name": "924734f72ba94a2285ece12dd9b66cda.png", "path": "poster_ocr_1024/924734f72ba94a2285ece12dd9b66cda.png", "path_original": "poster_ocr/924734f72ba94a2285ece12dd9b66cda.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['BUSINESSINSPIRATION DELICATIONS', 'Welcome', 'To Electro', 'To construcon project is compl Truc is complentcts a htyrt due to totrh nd otherentctsand otheron project is complentcts and otherentcts and other welcoem to maim.', '$10Entry', 'Mix', 'Call us :+098 123 4556 234', 'Business Designs']", "size": [ 1275, 1875 ], "texts": [ "BUSINESSINSPIRATION DELICATIONS", "Welcome", "To Electro", "To construcon project is compl Truc is complentcts a htyrt due to totrh nd otherentctsand otheron project is complentcts and otherentcts and other welcoem to maim.", "$10Entry", "Mix", "Call us :+098 123 4556 234", "Business Designs" ], "text_bbox": [ [ 109, 1470, 1016, 1618 ], [ 281, 642, 992, 762 ], [ 236, 809, 1018, 903 ], [ 105, 1699, 1156, 1761 ], [ 481, 295, 770, 504 ], [ 506, 975, 770, 1103 ], [ 831, 123, 1207, 208 ], [ 108, 1635, 584, 1701 ] ], "bbox_areas": [ 134236, 85320, 73508, 65162, 60401, 33792, 31960, 31416 ] }, { "task": "poster detection", "subtask": "", "name": "9152af0b31904f91a052be98af7e3db7.png", "path": "poster_ocr_1024/9152af0b31904f91a052be98af7e3db7.png", "path_original": "poster_ocr/9152af0b31904f91a052be98af7e3db7.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['SOCCER', 'CHAMPIONSHIP', 'SEMIFINALS', '10$ ENTRY | 2 FREE DRINKS | FREE PARKING', 'SATURDAY JULY 5TH 19PM', '- THE -', 'AT ZEPPELIN SPORTS BAR', '5$ BEER, HD SCREENS, 21+ TO ENTER', 'WWW.WEBSITE.COM ', 'FACEBOOK/YOURCLUB TWITTER/YOURCLUB YOUTUBE/YOURCLUB', 'ADRESS: 2231 MAIN STREET YOUR CITY CA.', 'VS', 'BRAZIL', '5544 445 455', 'SPAIN', 'FOR MORE INFO CALL US AT']", "size": [ 2700, 3450 ], "texts": [ "SOCCER", "CHAMPIONSHIP", "SEMIFINALS", "10$ ENTRY | 2 FREE DRINKS | FREE PARKING", "SATURDAY JULY 5TH 19PM", "- THE -", "AT ZEPPELIN SPORTS BAR", "5$ BEER, HD SCREENS, 21+ TO ENTER", "WWW.WEBSITE.COM ", "FACEBOOK/YOURCLUB TWITTER/YOURCLUB YOUTUBE/YOURCLUB", "ADRESS: 2231 MAIN STREET YOUR CITY CA.", "VS", "BRAZIL", "5544 445 455", "SPAIN", "FOR MORE INFO CALL US AT" ], "text_bbox": [ [ 537, 544, 2170, 873 ], [ 695, 919, 2009, 1159 ], [ 839, 2184, 1850, 2431 ], [ 694, 2496, 2002, 2593 ], [ 711, 1997, 1989, 2092 ], [ 1097, 295, 1602, 527 ], [ 692, 1300, 1997, 1376 ], [ 696, 2595, 2002, 2665 ], [ 978, 2727, 1715, 2830 ], [ 298, 3294, 2401, 3329 ], [ 759, 3175, 1939, 3221 ], [ 1205, 1599, 1486, 1769 ], [ 665, 1612, 1000, 1752 ], [ 1051, 3069, 1646, 3144 ], [ 1721, 1610, 2013, 1750 ], [ 1051, 3023, 1646, 3057 ] ], "bbox_areas": [ 537257, 315360, 249717, 126876, 121410, 117160, 99180, 91420, 75911, 73605, 54280, 47770, 46900, 44625, 40880, 20230 ] }, { "task": "poster detection", "subtask": "", "name": "90b9a2237ab5460b85e13b9bc40ca44f.png", "path": "poster_ocr_1024/90b9a2237ab5460b85e13b9bc40ca44f.png", "path_original": "poster_ocr/90b9a2237ab5460b85e13b9bc40ca44f.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['sundays', '$25 cocktail pitchers atthe Salsana waterside garden', 'f t v w', 'sizzlin hot', 'get more out of adobe photoshop with flyerheroes templates', 'www.flyerheroes.com', '$25', 'find our promotions on popular social networks', 'pitchers', 'ever sunday!']", "size": [ 1819, 2551 ], "texts": [ "sundays", "$25 cocktail pitchers atthe Salsana waterside garden", "f t v w", "sizzlin hot", "get more out of adobe photoshop with flyerheroes templates", "www.flyerheroes.com", "$25", "find our promotions on popular social networks", "pitchers", "ever sunday!" ], "text_bbox": [ [ 168, 1764, 1658, 2036 ], [ 269, 2091, 1546, 2192 ], [ 407, 2293, 1395, 2340 ], [ 466, 1681, 1361, 1725 ], [ 225, 2468, 1586, 2490 ], [ 590, 2396, 1223, 2427 ], [ 1288, 1321, 1474, 1419 ], [ 501, 2249, 1301, 2268 ], [ 1287, 1421, 1494, 1478 ], [ 1291, 1476, 1487, 1517 ] ], "bbox_areas": [ 405280, 128977, 46436, 39380, 29942, 19623, 18228, 15200, 11799, 8036 ] }, { "task": "poster detection", "subtask": "", "name": "9054964ab2734d2d978ede1f0bba35bb.png", "path": "poster_ocr_1024/9054964ab2734d2d978ede1f0bba35bb.png", "path_original": "poster_ocr/9054964ab2734d2d978ede1f0bba35bb.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['Slim', 'PLAN']", "size": [ 1080, 1920 ], "texts": [ "Slim", "PLAN" ], "text_bbox": [ [ 611, 62, 1040, 447 ], [ 829, 364, 1031, 406 ] ], "bbox_areas": [ 165165, 8484 ] }, { "task": "poster detection", "subtask": "", "name": "8ef586f2055f44ed90eaa44a95faadab.png", "path": "poster_ocr_1024/8ef586f2055f44ed90eaa44a95faadab.png", "path_original": "poster_ocr/8ef586f2055f44ed90eaa44a95faadab.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['MakeUp', 'School', 'Basic Level Course', '775 north, Central mall. Miami Florida', 'www.makeupschool.com', '14th', '+001 787 55 89', '2024', 'March', '10:00 am', 'contacts:']", "size": [ 2340, 3309 ], "texts": [ "MakeUp", "School", "Basic Level Course", "775 north, Central mall. Miami Florida", "www.makeupschool.com", "14th", "+001 787 55 89", "2024", "March", "10:00 am", "contacts:" ], "text_bbox": [ [ 372, 1839, 1975, 2297 ], [ 394, 2296, 1650, 2644 ], [ 532, 2715, 1743, 2852 ], [ 373, 3014, 1929, 3099 ], [ 703, 3113, 1605, 3231 ], [ 1702, 165, 2159, 366 ], [ 138, 282, 760, 360 ], [ 242, 1108, 597, 1240 ], [ 257, 959, 588, 1091 ], [ 1658, 444, 2209, 515 ], [ 141, 140, 517, 214 ] ], "bbox_areas": [ 734174, 437088, 165907, 132260, 106436, 91857, 48516, 46860, 43692, 39121, 27824 ] }, { "task": "poster detection", "subtask": "", "name": "8e38183e6ec145b5a2447f88b47c2bbe.png", "path": "poster_ocr_1024/8e38183e6ec145b5a2447f88b47c2bbe.png", "path_original": "poster_ocr/8e38183e6ec145b5a2447f88b47c2bbe.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['OneStep', 'One Step Night CLub Presents', 'Event @ 9pm | Drinks $3 | Girls Free, Till 12am', 'DJ SHEMUL, DJ JAHID', '24th August, 2013', '123 Street, Melbourne, Australia', 'Free Entry', 'www.yourwebsite.com', 'Music By']", "size": [ 1275, 1875 ], "texts": [ "OneStep", "One Step Night CLub Presents", "Event @ 9pm | Drinks $3 | Girls Free, Till 12am", "DJ SHEMUL, DJ JAHID", "24th August, 2013", "123 Street, Melbourne, Australia", "Free Entry", "www.yourwebsite.com", "Music By" ], "text_bbox": [ [ 536, 519, 976, 793 ], [ 203, 266, 638, 409 ], [ 362, 1456, 1081, 1502 ], [ 654, 1317, 1081, 1380 ], [ 192, 277, 405, 360 ], [ 664, 1507, 1082, 1542 ], [ 783, 850, 966, 892 ], [ 817, 1550, 1082, 1577 ], [ 972, 1272, 1082, 1302 ] ], "bbox_areas": [ 120560, 62205, 33074, 26901, 17679, 14630, 7686, 7155, 3300 ] }, { "task": "poster detection", "subtask": "", "name": "8d170bce6e6549a48e74edfcc818aa86.png", "path": "poster_ocr_1024/8d170bce6e6549a48e74edfcc818aa86.png", "path_original": "poster_ocr/8d170bce6e6549a48e74edfcc818aa86.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['IntroductionPortfolio ReviewPractise PartFood + Drink', 'Lifestyle PhWorkshop', 'STUDIO CROSS - MOUNT COOK ST 173, MELBOURNE', 'FROM 9 AM TO 6 PM']", "size": [ 1080, 1080 ], "texts": [ "IntroductionPortfolio ReviewPractise PartFood + Drink", "Lifestyle PhWorkshop", "STUDIO CROSS - MOUNT COOK ST 173, MELBOURNE", "FROM 9 AM TO 6 PM" ], "text_bbox": [ [ 335, 427, 743, 745 ], [ 306, 93, 767, 294 ], [ 115, 960, 963, 983 ], [ 374, 902, 703, 923 ] ], "bbox_areas": [ 129744, 92661, 19504, 6909 ] }, { "task": "poster detection", "subtask": "", "name": "8ccf48e225ac4a1092babcf9232b9343.png", "path": "poster_ocr_1024/8ccf48e225ac4a1092babcf9232b9343.png", "path_original": "poster_ocr/8ccf48e225ac4a1092babcf9232b9343.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['Sundayafternoon', '23', 'fashion InstagramStories Template', 'DAILY LOOK']", "size": [ 2250, 4000 ], "texts": [ "Sundayafternoon", "23", "fashion InstagramStories Template", "DAILY LOOK" ], "text_bbox": [ [ 896, 310, 2058, 818 ], [ 154, 3388, 797, 3861 ], [ 151, 155, 757, 300 ], [ 2001, 2702, 2047, 3273 ] ], "bbox_areas": [ 590296, 304139, 87870, 26266 ] }, { "task": "poster detection", "subtask": "", "name": "8bbf9cdf84884723b3aa06f7f9a12e44.png", "path": "poster_ocr_1024/8bbf9cdf84884723b3aa06f7f9a12e44.png", "path_original": "poster_ocr/8bbf9cdf84884723b3aa06f7f9a12e44.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['Girls', 'Girls', 'night', 'night', 'SAT JAN 12', 'dj master | dj envato | dj mixmaster', 'CLUB GRAPHICRIVER . 1235 WEST ENVATO RD. SAN BERNANDINO CA,', 'GIRLS', 'facebook .com/party', 'twitter .com/party', 'all night', 'enter free']", "size": [ 1275, 1875 ], "texts": [ "Girls", "Girls", "night", "night", "SAT JAN 12", "dj master | dj envato | dj mixmaster", "CLUB GRAPHICRIVER . 1235 WEST ENVATO RD. SAN BERNANDINO CA,", "GIRLS", "facebook .com/party", "twitter .com/party", "all night", "enter free" ], "text_bbox": [ [ 197, 849, 1111, 1097 ], [ 197, 849, 1111, 1097 ], [ 213, 1114, 1130, 1343 ], [ 213, 1114, 1130, 1343 ], [ 400, 1325, 918, 1506 ], [ 158, 1599, 1116, 1673 ], [ 190, 1748, 1085, 1770 ], [ 907, 291, 1064, 413 ], [ 712, 1686, 1116, 1728 ], [ 227, 1685, 604, 1727 ], [ 867, 375, 1018, 475 ], [ 890, 345, 1033, 443 ] ], "bbox_areas": [ 226672, 226672, 209993, 209993, 93758, 70892, 19690, 19154, 16968, 15834, 15100, 14014 ] }, { "task": "poster detection", "subtask": "", "name": "8b2a8a666ddf420185286189368ce727.png", "path": "poster_ocr_1024/8b2a8a666ddf420185286189368ce727.png", "path_original": "poster_ocr/8b2a8a666ddf420185286189368ce727.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['1 cup fresh spinach1/2 cup plain nonfat Greek yogurt1 cup blueberries1 1/2 cup water', 'Fruitalicious', '*New post on the blog.', 'YUMMMY!']", "size": [ 1080, 1920 ], "texts": [ "1 cup fresh spinach1/2 cup plain nonfat Greek yogurt1 cup blueberries1 1/2 cup water", "Fruitalicious", "*New post on the blog.", "YUMMMY!" ], "text_bbox": [ [ 92, 933, 989, 1281 ], [ 103, 529, 972, 885 ], [ 354, 1646, 725, 1675 ], [ 426, 1552, 652, 1578 ] ], "bbox_areas": [ 312156, 309364, 10759, 5876 ] }, { "task": "poster detection", "subtask": "", "name": "8b167c791aa542bd9a5efcab358a7a5f.png", "path": "poster_ocr_1024/8b167c791aa542bd9a5efcab358a7a5f.png", "path_original": "poster_ocr/8b167c791aa542bd9a5efcab358a7a5f.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['BE-lievein yourself', '52', 'Fashion InstagramStories Template', 'YOURWEB.COM']", "size": [ 2250, 4000 ], "texts": [ "BE-lievein yourself", "52", "Fashion InstagramStories Template", "YOURWEB.COM" ], "text_bbox": [ [ 149, 1697, 1145, 2127 ], [ 1522, 2219, 2154, 2712 ], [ 147, 2193, 770, 2338 ], [ 2000, 297, 2049, 1050 ] ], "bbox_areas": [ 428280, 311576, 90335, 36897 ] }, { "task": "poster detection", "subtask": "", "name": "8aa25162c1544d2085a325bf11068ba9.png", "path": "poster_ocr_1024/8aa25162c1544d2085a325bf11068ba9.png", "path_original": "poster_ocr/8aa25162c1544d2085a325bf11068ba9.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['DARKpharoe', 'dj BLACK | Dj MONEY | DJ SKULL', '9 PM . ENTRY 20$ . DRESS CODE . ALL IN BLACK', '-OVNI EVENT PRESENTs-', 'NightClub', 'FRIDAY 15th JULY', 'f facebook.com/YOURNAME', 'T twitter.com/YOURNAME']", "size": [ 1275, 1875 ], "texts": [ "DARKpharoe", "dj BLACK | Dj MONEY | DJ SKULL", "9 PM . ENTRY 20$ . DRESS CODE . ALL IN BLACK", "-OVNI EVENT PRESENTs-", "NightClub", "FRIDAY 15th JULY", "f facebook.com/YOURNAME", "T twitter.com/YOURNAME" ], "text_bbox": [ [ 149, 1197, 1127, 1501 ], [ 189, 1529, 1087, 1587 ], [ 195, 1606, 1082, 1639 ], [ 211, 1129, 1064, 1159 ], [ 510, 1663, 778, 1724 ], [ 466, 1062, 808, 1107 ], [ 203, 1663, 506, 1695 ], [ 789, 1663, 1069, 1695 ] ], "bbox_areas": [ 297312, 52084, 29271, 25590, 16348, 15390, 9696, 8960 ] }, { "task": "poster detection", "subtask": "", "name": "8a39570a93ef4b04b7853f941fc891b8.png", "path": "poster_ocr_1024/8a39570a93ef4b04b7853f941fc891b8.png", "path_original": "poster_ocr/8a39570a93ef4b04b7853f941fc891b8.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['WorldForest Day', 'Lorem ipsum dolor sit amet, consectetur adipiscing elit sed do eiusmod tempor incididunt ut labore et dolore.', 'Forest protection now', '21 March', 'www.yoursite.com', 'your account', 'your account', 'your account']", "size": [ 1753, 2480 ], "texts": [ "WorldForest Day", "Lorem ipsum dolor sit amet, consectetur adipiscing elit sed do eiusmod tempor incididunt ut labore et dolore.", "Forest protection now", "21 March", "www.yoursite.com", "your account", "your account", "your account" ], "text_bbox": [ [ 456, 1124, 1395, 1485 ], [ 401, 1614, 1395, 1803 ], [ 416, 1419, 1358, 1600 ], [ 1382, 2233, 1545, 2345 ], [ 674, 1933, 1082, 1958 ], [ 1129, 2320, 1310, 2348 ], [ 818, 2322, 998, 2350 ], [ 501, 2325, 681, 2353 ] ], "bbox_areas": [ 338979, 187866, 170502, 18256, 10200, 5068, 5040, 5040 ] }, { "task": "poster detection", "subtask": "", "name": "8925471c05d84fdeb8e890bf6961ebd5.png", "path": "poster_ocr_1024/8925471c05d84fdeb8e890bf6961ebd5.png", "path_original": "poster_ocr/8925471c05d84fdeb8e890bf6961ebd5.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['NewDresses.', 'All of the latest styles are just a click away', 'shop new arrivals']", "size": [ 750, 1050 ], "texts": [ "NewDresses.", "All of the latest styles are just a click away", "shop new arrivals" ], "text_bbox": [ [ 102, 378, 633, 613 ], [ 103, 661, 629, 688 ], [ 135, 747, 446, 765 ] ], "bbox_areas": [ 124785, 14202, 5598 ] }, { "task": "poster detection", "subtask": "", "name": "88402616164a4fca96e1ca7a5d3f80a1.png", "path": "poster_ocr_1024/88402616164a4fca96e1ca7a5d3f80a1.png", "path_original": "poster_ocr/88402616164a4fca96e1ca7a5d3f80a1.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['bikinis', '50%offsale', 'dream deals just for you this summetime', 'upgrade your']", "size": [ 750, 1050 ], "texts": [ "bikinis", "50%offsale", "dream deals just for you this summetime", "upgrade your" ], "text_bbox": [ [ 38, 105, 707, 275 ], [ 516, 434, 654, 603 ], [ 120, 295, 626, 319 ], [ 215, 65, 530, 88 ] ], "bbox_areas": [ 113730, 23322, 12144, 7245 ] }, { "task": "poster detection", "subtask": "", "name": "86c30b9e8e2d456fb50a628cec91c70a.png", "path": "poster_ocr_1024/86c30b9e8e2d456fb50a628cec91c70a.png", "path_original": "poster_ocr/86c30b9e8e2d456fb50a628cec91c70a.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['GIRLS', 'Night', 'dj LOVER | Dj CLASS | DJ GRAPHICS', 'GoldClub', 'ADMISSION 10$ | FREE FOR LADIES | START 7:30 PM', 'YOUR ENTERTAINMENT PRESENTS', '25.10.2015', 'F facebook.com/YOURNAME', 'T twitter.com/YOURNAME', 'W wWw.website.com']", "size": [ 1275, 1875 ], "texts": [ "GIRLS", "Night", "dj LOVER | Dj CLASS | DJ GRAPHICS", "GoldClub", "ADMISSION 10$ | FREE FOR LADIES | START 7:30 PM", "YOUR ENTERTAINMENT PRESENTS", "25.10.2015", "F facebook.com/YOURNAME", "T twitter.com/YOURNAME", "W wWw.website.com" ], "text_bbox": [ [ 236, 291, 1040, 709 ], [ 502, 710, 1042, 892 ], [ 236, 1543, 1039, 1609 ], [ 485, 1454, 793, 1533 ], [ 248, 1630, 1022, 1657 ], [ 260, 229, 1011, 255 ], [ 513, 151, 763, 199 ], [ 262, 1657, 512, 1682 ], [ 547, 1657, 779, 1682 ], [ 815, 1657, 1009, 1681 ] ], "bbox_areas": [ 336072, 98280, 52998, 24332, 20898, 19526, 12000, 6250, 5800, 4656 ] }, { "task": "poster detection", "subtask": "", "name": "85c022ca5e33445aafee198919e268b5.png", "path": "poster_ocr_1024/85c022ca5e33445aafee198919e268b5.png", "path_original": "poster_ocr/85c022ca5e33445aafee198919e268b5.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['SILVERDRINK', 'dj BLACK | Dj MONEY | DJ STRIKE', '9 PM || ENTRY 20$ || DRESS CODE : ALL IN BLACK', '-OVNI EVENT PRESENTs-', 'NightClub', 'FRIDAY 15th JULY', 'f facebook.com/YOURNAME', 'T twitter.com/YOURNAME']", "size": [ 1275, 1875 ], "texts": [ "SILVERDRINK", "dj BLACK | Dj MONEY | DJ STRIKE", "9 PM || ENTRY 20$ || DRESS CODE : ALL IN BLACK", "-OVNI EVENT PRESENTs-", "NightClub", "FRIDAY 15th JULY", "f facebook.com/YOURNAME", "T twitter.com/YOURNAME" ], "text_bbox": [ [ 173, 174, 1103, 464 ], [ 217, 1491, 1064, 1547 ], [ 233, 1569, 1052, 1604 ], [ 210, 110, 1063, 141 ], [ 523, 1621, 773, 1680 ], [ 486, 1422, 806, 1465 ], [ 234, 1622, 518, 1652 ], [ 784, 1621, 1045, 1651 ] ], "bbox_areas": [ 269700, 47432, 28665, 26443, 14750, 13760, 8520, 7830 ] }, { "task": "poster detection", "subtask": "", "name": "8562918161344ed787afd730933e8ee1.png", "path": "poster_ocr_1024/8562918161344ed787afd730933e8ee1.png", "path_original": "poster_ocr/8562918161344ed787afd730933e8ee1.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['%30', \"Lorem Ipsum is simply dummy text of the printing and typesetting industry. Lorem Ipsum has been the industry's standard dummy text ever since the 1500s, when an unknown printer took a galley of type and scrambled it to make a type specimen book. It has survived not only five centuries, but also the leap into\", 'Form March 28', 'www.yourwebsite.com', 'To April 3', 'sale']", "size": [ 2555, 3583 ], "texts": [ "%30", "Lorem Ipsum is simply dummy text of the printing and typesetting industry. Lorem Ipsum has been the industry's standard dummy text ever since the 1500s, when an unknown printer took a galley of type and scrambled it to make a type specimen book. It has survived not only five centuries, but also the leap into", "Form March 28", "www.yourwebsite.com", "To April 3", "sale" ], "text_bbox": [ [ 293, 1011, 2261, 2037 ], [ 336, 2445, 2219, 2686 ], [ 318, 2156, 1127, 2268 ], [ 911, 2952, 1644, 3048 ], [ 1368, 2156, 1900, 2268 ], [ 289, 781, 570, 908 ] ], "bbox_areas": [ 2019168, 453803, 90608, 70368, 59584, 35687 ] }, { "task": "poster detection", "subtask": "", "name": "84adb8365f0541c38fa2a1878234679c.png", "path": "poster_ocr_1024/84adb8365f0541c38fa2a1878234679c.png", "path_original": "poster_ocr/84adb8365f0541c38fa2a1878234679c.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['CHEEKYYELLOW', 'PARTY', '29.06.12', 'WWW.YOURWEBSITE.CO.UK | EMAIL@YOUREVENT.CO.UK | 0123456789', 'YOUR NAMEPRESENTS']", "size": [ 1275, 1875 ], "texts": [ "CHEEKYYELLOW", "PARTY", "29.06.12", "WWW.YOURWEBSITE.CO.UK | EMAIL@YOUREVENT.CO.UK | 0123456789", "YOUR NAMEPRESENTS" ], "text_bbox": [ [ 199, 1211, 1075, 1560 ], [ 193, 1594, 1082, 1630 ], [ 770, 721, 1088, 788 ], [ 292, 1694, 982, 1724 ], [ 943, 194, 1134, 295 ] ], "bbox_areas": [ 305724, 32004, 21306, 20700, 19291 ] }, { "task": "poster detection", "subtask": "", "name": "8206435d61664df2b09f35366381fbea.png", "path": "poster_ocr_1024/8206435d61664df2b09f35366381fbea.png", "path_original": "poster_ocr/8206435d61664df2b09f35366381fbea.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['from the dust', 'the s.i.g.i.t', 'hydro parklondon', 'Visit www.yoursite.com for more infoFollow us on facebook & twitter', '17.jun.2013', 'Alternative Live Concert', 'And other special guests', 'The moonlights', 'Kings of leon', 'Shelby67 Presents', 'Show will be live from', 'doors open at 10.PM', 'Featuring', '5$', 'tickets ->']", "size": [ 2625, 3375 ], "texts": [ "from the dust", "the s.i.g.i.t", "hydro parklondon", "Visit www.yoursite.com for more infoFollow us on facebook & twitter", "17.jun.2013", "Alternative Live Concert", "And other special guests", "The moonlights", "Kings of leon", "Shelby67 Presents", "Show will be live from", "doors open at 10.PM", "Featuring", "5$", "tickets ->" ], "text_bbox": [ [ 186, 261, 2417, 761 ], [ 192, 1053, 1402, 1397 ], [ 1478, 2436, 2396, 2801 ], [ 314, 2844, 1524, 3079 ], [ 1373, 2100, 2353, 2377 ], [ 215, 688, 1458, 888 ], [ 243, 1633, 1349, 1798 ], [ 224, 1510, 1069, 1678 ], [ 214, 1390, 949, 1554 ], [ 889, 176, 1685, 324 ], [ 1521, 2346, 2372, 2473 ], [ 576, 2566, 985, 2757 ], [ 178, 1012, 589, 1112 ], [ 2103, 2922, 2290, 3105 ], [ 1606, 2972, 1995, 3048 ] ], "bbox_areas": [ 1115500, 416240, 335070, 284350, 271460, 248600, 182490, 141960, 120540, 117808, 108077, 78119, 41100, 34221, 29564 ] }, { "task": "poster detection", "subtask": "", "name": "813d677cbdf04c24b182dfc1de572019.png", "path": "poster_ocr_1024/813d677cbdf04c24b182dfc1de572019.png", "path_original": "poster_ocr/813d677cbdf04c24b182dfc1de572019.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['thesummertechnoparty2013', 'sunday june 30th', 'doors open at 9pm', 'envatoentertainment', 'envato dance club 95 graphicriver street', 'freeenter', 'www.envatodanceclub.com']", "size": [ 1275, 1875 ], "texts": [ "thesummertechnoparty2013", "sunday june 30th", "doors open at 9pm", "envatoentertainment", "envato dance club 95 graphicriver street", "freeenter", "www.envatodanceclub.com" ], "text_bbox": [ [ 403, 534, 873, 1342 ], [ 351, 1509, 925, 1572 ], [ 351, 1581, 925, 1644 ], [ 390, 120, 886, 158 ], [ 352, 1654, 925, 1676 ], [ 243, 565, 374, 661 ], [ 351, 1686, 925, 1702 ] ], "bbox_areas": [ 379760, 36162, 36162, 18848, 12606, 12576, 9184 ] }, { "task": "poster detection", "subtask": "", "name": "802352fa7c8341da8291d27873a02a6f.png", "path": "poster_ocr_1024/802352fa7c8341da8291d27873a02a6f.png", "path_original": "poster_ocr/802352fa7c8341da8291d27873a02a6f.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['LifestyleStories', 'New posts onmy blog', 'Creato', 'www.creatogallery.com', 'Vizit My Blog']", "size": [ 1200, 1200 ], "texts": [ "LifestyleStories", "New posts onmy blog", "Creato", "www.creatogallery.com", "Vizit My Blog" ], "text_bbox": [ [ 250, 377, 781, 606 ], [ 327, 689, 481, 734 ], [ 250, 302, 366, 350 ], [ 100, 1069, 270, 1083 ], [ 499, 1069, 623, 1087 ] ], "bbox_areas": [ 121599, 6930, 5568, 2380, 2232 ] }, { "task": "poster detection", "subtask": "", "name": "7b830ef82386451caa7770fdb4647a8b.png", "path": "poster_ocr_1024/7b830ef82386451caa7770fdb4647a8b.png", "path_original": "poster_ocr/7b830ef82386451caa7770fdb4647a8b.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['Bible Study', 'DECEMBER 25TH 9PM TO12PM ', 'BESTCHURCHSTUDYIDEA...', 'RESPONSIBLES', 'Lorem Ipsum is simply dummy text of the printing and typesetting industry. galley of type and scrambled it to make special orem Ipsum is simply dummy text of the prin.', 'THEN YOU WILL KNOW THE TRUTH', 'BEST CHURCH THEME', 'BEST CONCERT EVER', 'ANAYA22 PRSENTS']", "size": [ 1275, 1875 ], "texts": [ "Bible Study", "DECEMBER 25TH 9PM TO12PM ", "BESTCHURCHSTUDYIDEA...", "RESPONSIBLES", "Lorem Ipsum is simply dummy text of the printing and typesetting industry. galley of type and scrambled it to make special orem Ipsum is simply dummy text of the prin.", "THEN YOU WILL KNOW THE TRUTH", "BEST CHURCH THEME", "BEST CONCERT EVER", "ANAYA22 PRSENTS" ], "text_bbox": [ [ 79, 176, 1195, 389 ], [ 80, 1537, 1012, 1612 ], [ 886, 1127, 1165, 1352 ], [ 109, 1279, 823, 1356 ], [ 79, 1628, 1078, 1683 ], [ 78, 408, 933, 471 ], [ 80, 1443, 727, 1518 ], [ 498, 1746, 774, 1766 ], [ 504, 102, 764, 122 ] ], "bbox_areas": [ 237708, 69900, 62775, 54978, 54945, 53865, 48525, 5520, 5200 ] }, { "task": "poster detection", "subtask": "", "name": "7b59e50490b249c18ff6277155c3fefd.png", "path": "poster_ocr_1024/7b59e50490b249c18ff6277155c3fefd.png", "path_original": "poster_ocr/7b59e50490b249c18ff6277155c3fefd.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['SUMMER', 'Beach Party', 'SATURDAY AUGUST 15TH', '- MONTAUK BEACH PARTY STARTS AT 1PM -', '35$ ENTRY / BIKINI CONTEST / FREE PARKING', '21+ TO ENTER FOR MORE INFO CALL 5554 445 445', 'WWW.WEBSITE.COM', '0921 MAIN STREET YOUR CITY NY.']", "size": [ 2700, 3450 ], "texts": [ "SUMMER", "Beach Party", "SATURDAY AUGUST 15TH", "- MONTAUK BEACH PARTY STARTS AT 1PM -", "35$ ENTRY / BIKINI CONTEST / FREE PARKING", "21+ TO ENTER FOR MORE INFO CALL 5554 445 445", "WWW.WEBSITE.COM", "0921 MAIN STREET YOUR CITY NY." ], "text_bbox": [ [ 278, 1077, 2389, 1888 ], [ 560, 1593, 2175, 2020 ], [ 403, 2000, 2301, 2223 ], [ 595, 2265, 2111, 2431 ], [ 536, 2500, 2226, 2639 ], [ 538, 2588, 2234, 2722 ], [ 930, 2786, 1865, 2958 ], [ 988, 2948, 1828, 3027 ] ], "bbox_areas": [ 1712021, 689605, 423254, 251656, 234910, 227264, 160820, 66360 ] }, { "task": "poster detection", "subtask": "", "name": "7b500d59be644696985e46c7e391b87b.png", "path": "poster_ocr_1024/7b500d59be644696985e46c7e391b87b.png", "path_original": "poster_ocr/7b500d59be644696985e46c7e391b87b.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['love', 'elec', 'tro', 'this autumnmost awaited event this yearbest deejaysmolestie a, ultricies porta urna. pat a, convallisfor more information:yourparty.comvelit, rhoncus eu, luctus et interdum adipiscing wisi. Aliquam erat ac ipsum. Integer aliquam purus. Quisque lorem tortor fringilla sed, vestibulum id, eleifend justo vel bibendum sapien massa ac turpis faucibus orci luctus non, consectetuer lobortis quis, varius in, purus. Integer ultrices posuere cubilia Curae, Nulla ipsum dolor lacus, suscipit adipiscing. Cum sociis natoque penatibus et ultrices volutpat. Nullam wisi ultricies a, gravida vitae, dapibus risus ante sodales lectus blandit eu, tempor diam pede cursus vitae, ultricies eu, faucibus quis, porttitor eros cursus lectus, pellentesque eget, bibendum a, gravida ullamcorper quam. Nullam viverra consectetuer. Quisque cursus et, porttitor risus. Aliquam sem. In hendrerit nulla quam nunc, accumsan congue. Lorem ipsum primis in nibh vel risus. Sed vel lectus. Ut sagittis, ipsum dolor quam.', '09.03', '22.11.201322:00CLUB NAMECITYGIRLS: 30$BOYS: 50$', 'Lorem ipsum pendisse a pelle', 'yourparty.com', '------------------------------------------------', '---------------------------------------------------------------------------------------', '------------------------------------', '------------------------']", "size": [ 2550, 3570 ], "texts": [ "love", "elec", "tro", "this autumnmost awaited event this yearbest deejaysmolestie a, ultricies porta urna. pat a, convallisfor more information:yourparty.comvelit, rhoncus eu, luctus et interdum adipiscing wisi. Aliquam erat ac ipsum. Integer aliquam purus. Quisque lorem tortor fringilla sed, vestibulum id, eleifend justo vel bibendum sapien massa ac turpis faucibus orci luctus non, consectetuer lobortis quis, varius in, purus. Integer ultrices posuere cubilia Curae, Nulla ipsum dolor lacus, suscipit adipiscing. Cum sociis natoque penatibus et ultrices volutpat. Nullam wisi ultricies a, gravida vitae, dapibus risus ante sodales lectus blandit eu, tempor diam pede cursus vitae, ultricies eu, faucibus quis, porttitor eros cursus lectus, pellentesque eget, bibendum a, gravida ullamcorper quam. Nullam viverra consectetuer. Quisque cursus et, porttitor risus. Aliquam sem. In hendrerit nulla quam nunc, accumsan congue. Lorem ipsum primis in nibh vel risus. Sed vel lectus. Ut sagittis, ipsum dolor quam.", "09.03", "22.11.201322:00CLUB NAMECITYGIRLS: 30$BOYS: 50$", "Lorem ipsum pendisse a pelle", "yourparty.com", "------------------------------------------------", "---------------------------------------------------------------------------------------", "------------------------------------", "------------------------" ], "text_bbox": [ [ 399, 358, 1409, 700 ], [ 396, 733, 1295, 1075 ], [ 411, 1110, 1120, 1451 ], [ 399, 2367, 1146, 2690 ], [ 392, 1937, 913, 2126 ], [ 2066, 255, 2307, 581 ], [ 397, 2223, 1080, 2254 ], [ 409, 2770, 775, 2801 ], [ 784, 1553, 786, 2293 ], [ 0, 356, 707, 357 ], [ 1555, 359, 2109, 360 ], [ 1279, 1451, 1647, 1452 ] ], "bbox_areas": [ 345420, 307458, 241769, 241281, 98469, 78566, 21173, 11346, 1480, 707, 554, 368 ] }, { "task": "poster detection", "subtask": "", "name": "749bd6e0725a4472bcd96fcdbebc16b1.png", "path": "poster_ocr_1024/749bd6e0725a4472bcd96fcdbebc16b1.png", "path_original": "poster_ocr/749bd6e0725a4472bcd96fcdbebc16b1.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['ROLLER', 'RUMBLE SHOWDOWN', 'DOORS AT SEVEN FIRST BOUT STARTS AT NINEONE NIGHT ONLY 24TH OCTOBER 2012FREE PUNK ROCK GIG - AN ALL AGES EVENT', 'BOMB CitIBOMBSHELLS', 'RATT CitIROLLERGIRLS', 'BAYSIDE BLONDESHELLS PRESENT', 'VS']", "size": [ 1275, 1875 ], "texts": [ "ROLLER", "RUMBLE SHOWDOWN", "DOORS AT SEVEN FIRST BOUT STARTS AT NINEONE NIGHT ONLY 24TH OCTOBER 2012FREE PUNK ROCK GIG - AN ALL AGES EVENT", "BOMB CitIBOMBSHELLS", "RATT CitIROLLERGIRLS", "BAYSIDE BLONDESHELLS PRESENT", "VS" ], "text_bbox": [ [ 103, 186, 1192, 500 ], [ 88, 485, 1169, 640 ], [ 127, 1607, 1160, 1716 ], [ 109, 1362, 459, 1474 ], [ 836, 1362, 1181, 1474 ], [ 104, 154, 857, 175 ], [ 587, 1370, 703, 1465 ] ], "bbox_areas": [ 341946, 167555, 112597, 39200, 38640, 15813, 11020 ] }, { "task": "poster detection", "subtask": "", "name": "74516666b4ed4838b9e410903ebf0d5d.png", "path": "poster_ocr_1024/74516666b4ed4838b9e410903ebf0d5d.png", "path_original": "poster_ocr/74516666b4ed4838b9e410903ebf0d5d.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['RETRO', 'ELECTRO', 'SATURDAY MAY 15TH', 'SUMMER JAM4', '>>>DJ DAZE DPPDJ SLICKSDJ PAUL', 'AT THE ZEPPELIN ', '>>>UNDERGROUND ELCTRO NIGHT', 'DOORS OPEN @9PM', 'FACEBOOK/EVENT TWITTER/EVENT YOUTUBE/EVENT', '5421 MAIN STREETYOUR CITY CA.', '4TH', '20$ ENTRY', 'ZEPPELIN & RADIO 54 PRESENT', 'FEATURING SPECIAL GUESTS', 'WWW.WEBSITE.COM', '5533 223 223', 'EDITION', '& MC PAULY']", "size": [ 2700, 3450 ], "texts": [ "RETRO", "ELECTRO", "SATURDAY MAY 15TH", "SUMMER JAM4", ">>>DJ DAZE DPPDJ SLICKSDJ PAUL", "AT THE ZEPPELIN ", ">>>UNDERGROUND ELCTRO NIGHT", "DOORS OPEN @9PM", "FACEBOOK/EVENT TWITTER/EVENT YOUTUBE/EVENT", "5421 MAIN STREETYOUR CITY CA.", "4TH", "20$ ENTRY", "ZEPPELIN & RADIO 54 PRESENT", "FEATURING SPECIAL GUESTS", "WWW.WEBSITE.COM", "5533 223 223", "EDITION", "& MC PAULY" ], "text_bbox": [ [ 359, 1821, 2357, 2266 ], [ 355, 2270, 1637, 2478 ], [ 348, 2534, 1887, 2660 ], [ 1986, 241, 2361, 534 ], [ 346, 305, 821, 532 ], [ 347, 2797, 1415, 2889 ], [ 348, 2944, 1198, 3024 ], [ 349, 3031, 1034, 3101 ], [ 792, 3294, 1905, 3327 ], [ 1959, 2998, 2364, 3076 ], [ 2023, 2304, 2251, 2427 ], [ 362, 3124, 751, 3193 ], [ 997, 125, 1704, 162 ], [ 346, 244, 1037, 281 ], [ 1882, 3161, 2363, 3198 ], [ 2057, 3117, 2364, 3153 ], [ 2023, 2432, 2253, 2474 ], [ 347, 543, 569, 574 ] ], "bbox_areas": [ 889110, 266656, 193914, 109875, 107825, 98256, 68000, 47950, 36729, 31590, 28044, 26841, 26159, 25567, 17797, 11052, 9660, 6882 ] }, { "task": "poster detection", "subtask": "", "name": "73fdeca96eef4d198f7a272b2af1d7f6.png", "path": "poster_ocr_1024/73fdeca96eef4d198f7a272b2af1d7f6.png", "path_original": "poster_ocr/73fdeca96eef4d198f7a272b2af1d7f6.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['ELECTRO', 'TAKEOVER', '1894 MAIN STREET, CITY, CA. NEAR THE PLAZA ', '20$ ENTRY FREE FOR GIRLS 5$ DRINKS VALET PARKING', '@ CLUB ZEPPELIN', 'DOORS OPEN AT 9PM', 'WWW.SUMMERFEST.COM', 'OR CALL US: 555 666 444', '2014', 'SATURDAY', 'SUMMERFEST', 'DJ SLICK D', 'DJ SHADE', 'JULY 14TH', 'DJ PAULY', 'ZEPPELIN PRESENTS', 'DJ VOX', 'EDITION', 'FACEBOOK', 'YOUTUBE', 'TWITTER', 'VIMEO']", "size": [ 2700, 3450 ], "texts": [ "ELECTRO", "TAKEOVER", "1894 MAIN STREET, CITY, CA. NEAR THE PLAZA ", "20$ ENTRY FREE FOR GIRLS 5$ DRINKS VALET PARKING", "@ CLUB ZEPPELIN", "DOORS OPEN AT 9PM", "WWW.SUMMERFEST.COM", "OR CALL US: 555 666 444", "2014", "SATURDAY", "SUMMERFEST", "DJ SLICK D", "DJ SHADE", "JULY 14TH", "DJ PAULY", "ZEPPELIN PRESENTS", "DJ VOX", "EDITION", "FACEBOOK", "YOUTUBE", "TWITTER", "VIMEO" ], "text_bbox": [ [ 857, 1523, 1841, 1738 ], [ 857, 1751, 1840, 1939 ], [ 498, 2779, 2204, 2849 ], [ 500, 2691, 2203, 2754 ], [ 929, 639, 1776, 726 ], [ 1033, 748, 1666, 797 ], [ 284, 3093, 961, 3136 ], [ 1764, 3095, 2417, 3138 ], [ 1214, 1253, 1484, 1353 ], [ 159, 1687, 565, 1750 ], [ 1120, 2100, 1575, 2156 ], [ 841, 330, 1242, 393 ], [ 1501, 326, 1888, 389 ], [ 2141, 1695, 2532, 1755 ], [ 284, 330, 659, 390 ], [ 1082, 160, 1619, 201 ], [ 2121, 327, 2415, 389 ], [ 1214, 1358, 1484, 1418 ], [ 366, 3254, 658, 3294 ], [ 2158, 3252, 2417, 3292 ], [ 1037, 3256, 1275, 3295 ], [ 1646, 3254, 1819, 3294 ] ], "bbox_areas": [ 211560, 184804, 119420, 107289, 73689, 31017, 29111, 28079, 27000, 25578, 25480, 25263, 24381, 23460, 22500, 22017, 18228, 16200, 11680, 10360, 9282, 6920 ] }, { "task": "poster detection", "subtask": "", "name": "6fb71b5134a446f994287f1408354e22.png", "path": "poster_ocr_1024/6fb71b5134a446f994287f1408354e22.png", "path_original": "poster_ocr/6fb71b5134a446f994287f1408354e22.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['our new summer lookbook is available on our websiteshop now', 'wild & free']", "size": [ 750, 1050 ], "texts": [ "our new summer lookbook is available on our websiteshop now", "wild & free" ], "text_bbox": [ [ 154, 801, 598, 937 ], [ 77, 92, 662, 177 ] ], "bbox_areas": [ 60384, 49725 ] }, { "task": "poster detection", "subtask": "", "name": "6f9c16ad7a0c46ef88c099b0ac131d76.png", "path": "poster_ocr_1024/6f9c16ad7a0c46ef88c099b0ac131d76.png", "path_original": "poster_ocr/6f9c16ad7a0c46ef88c099b0ac131d76.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['KARAOKE', 'DOORS OPENAT 8PM', 'AT CLUB ENVATO', 'MC DAVEY', '20$', 'DONNA J', 'SIR JESS', 'MAY 3RD', 'SUNDAY', '9442 MAIN STREET', 'ENTRY', '& SPECIAL GUEST', ' A GRAPHIC PRESENTS', 'YOUR CITY CA.', 'WWW.WEBSITE.COM', 'FEATURING:', '5542 552 552', 'CLUB@ADRESS.COM', 'FACEBOOK/YOURCLUB', 'YOUTUBE/YOURCLUB', 'TWITTER/YOURCLUB', 'VIMEO/YOURCLUB']", "size": [ 2700, 3450 ], "texts": [ "KARAOKE", "DOORS OPENAT 8PM", "AT CLUB ENVATO", "MC DAVEY", "20$", "DONNA J", "SIR JESS", "MAY 3RD", "SUNDAY", "9442 MAIN STREET", "ENTRY", "& SPECIAL GUEST", " A GRAPHIC PRESENTS", "YOUR CITY CA.", "WWW.WEBSITE.COM", "FEATURING:", "5542 552 552", "CLUB@ADRESS.COM", "FACEBOOK/YOURCLUB", "YOUTUBE/YOURCLUB", "TWITTER/YOURCLUB", "VIMEO/YOURCLUB" ], "text_bbox": [ [ 242, 2010, 2447, 2443 ], [ 1792, 2394, 2453, 2605 ], [ 226, 1958, 1210, 2092 ], [ 275, 1237, 918, 1325 ], [ 374, 635, 704, 795 ], [ 277, 1430, 825, 1519 ], [ 273, 1332, 814, 1422 ], [ 1878, 206, 2372, 297 ], [ 1873, 304, 2370, 392 ], [ 1775, 2897, 2423, 2947 ], [ 374, 796, 704, 890 ], [ 277, 1537, 875, 1588 ], [ 1019, 184, 1678, 228 ], [ 1923, 2951, 2424, 3001 ], [ 1878, 3120, 2425, 3160 ], [ 278, 1138, 618, 1181 ], [ 2087, 3034, 2425, 3070 ], [ 2038, 3084, 2426, 3114 ], [ 274, 3240, 677, 3267 ], [ 1600, 3241, 1979, 3268 ], [ 788, 3239, 1152, 3266 ], [ 2096, 3240, 2426, 3267 ] ], "bbox_areas": [ 954765, 139471, 131856, 56584, 52800, 48772, 48690, 44954, 43736, 32400, 31020, 30498, 28996, 25050, 21880, 14620, 12168, 11640, 10881, 10233, 9828, 8910 ] }, { "task": "poster detection", "subtask": "", "name": "6ef683e751084e2190439afee2072b77.png", "path": "poster_ocr_1024/6ef683e751084e2190439afee2072b77.png", "path_original": "poster_ocr/6ef683e751084e2190439afee2072b77.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['Let go oftoxic control,in order toregain healthycontrol.', '- Kayla Rose Kotecki -']", "size": [ 1080, 1920 ], "texts": [ "Let go oftoxic control,in order toregain healthycontrol.", "- Kayla Rose Kotecki -" ], "text_bbox": [ [ 136, 505, 947, 1361 ], [ 164, 1543, 909, 1577 ] ], "bbox_areas": [ 694216, 25330 ] }, { "task": "poster detection", "subtask": "", "name": "6df53f71f46d4b76b5c2681bb0c95dc8.png", "path": "poster_ocr_1024/6df53f71f46d4b76b5c2681bb0c95dc8.png", "path_original": "poster_ocr/6df53f71f46d4b76b5c2681bb0c95dc8.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['MusicSpecial night', 'BUSINESSINSPIRATION DELICA', 'To construcon project is compl Truc is compleyrt due to totrh nd otherentcts to the main text of thpartyand otheron project is complentcts and of the tentherentcts and other welcoem to maim.', 'BUSINESSINSPIRN DEICATINS', 'BUSINESSINSPIRN DEICATINS', 'Call us :+098 123 4556 234', '$11', 'DJ Low-J & Yolk-X', 'Per Shot']", "size": [ 1275, 1875 ], "texts": [ "MusicSpecial night", "BUSINESSINSPIRATION DELICA", "To construcon project is compl Truc is compleyrt due to totrh nd otherentcts to the main text of thpartyand otheron project is complentcts and of the tentherentcts and other welcoem to maim.", "BUSINESSINSPIRN DEICATINS", "BUSINESSINSPIRN DEICATINS", "Call us :+098 123 4556 234", "$11", "DJ Low-J & Yolk-X", "Per Shot" ], "text_bbox": [ [ 252, 776, 999, 1016 ], [ 274, 1452, 975, 1600 ], [ 261, 1633, 972, 1759 ], [ 338, 608, 925, 746 ], [ 365, 1056, 904, 1181 ], [ 98, 99, 474, 184 ], [ 971, 118, 1162, 230 ], [ 481, 1206, 827, 1237 ], [ 987, 241, 1174, 272 ] ], "bbox_areas": [ 179280, 103748, 89586, 81006, 67375, 31960, 21392, 10726, 5797 ] }, { "task": "poster detection", "subtask": "", "name": "6dc22400ff6148499fbd4ec51a3300de.png", "path": "poster_ocr_1024/6dc22400ff6148499fbd4ec51a3300de.png", "path_original": "poster_ocr/6dc22400ff6148499fbd4ec51a3300de.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['PURESOUND', 'FEATURINGDEEJAY ENVATO // DEEJAY GRAPHICS', 'YOURNAME PRESENTS', 'ELECTRO / HOUSE / DUBSTEP/ECLECTIC', 'SATURDAY 20TH AUGUST 2014', 'ENTRY 10$ free admission for ladies until 10', 'TICKET INFO AND LATEST NEWS ON WWW.WEBSITE.COM', 'START AT 9 PM', 'NightClub', 'www.facebook.com', 'www.twitter.com', 'www.yourwebsite.com']", "size": [ 1275, 1875 ], "texts": [ "PURESOUND", "FEATURINGDEEJAY ENVATO // DEEJAY GRAPHICS", "YOURNAME PRESENTS", "ELECTRO / HOUSE / DUBSTEP/ECLECTIC", "SATURDAY 20TH AUGUST 2014", "ENTRY 10$ free admission for ladies until 10", "TICKET INFO AND LATEST NEWS ON WWW.WEBSITE.COM", "START AT 9 PM", "NightClub", "www.facebook.com", "www.twitter.com", "www.yourwebsite.com" ], "text_bbox": [ [ 315, 710, 947, 1051 ], [ 238, 1353, 1017, 1450 ], [ 248, 200, 1010, 265 ], [ 237, 1452, 1018, 1493 ], [ 339, 296, 919, 345 ], [ 238, 1520, 1018, 1552 ], [ 237, 1559, 1017, 1581 ], [ 439, 364, 807, 402 ], [ 524, 1698, 746, 1749 ], [ 336, 1624, 599, 1646 ], [ 744, 1626, 963, 1648 ], [ 525, 1754, 745, 1775 ] ], "bbox_areas": [ 215512, 75563, 49530, 32021, 28420, 24960, 17160, 13984, 11322, 5786, 4818, 4620 ] }, { "task": "poster detection", "subtask": "", "name": "6da895a65f0a400491abf53a909785ce.png", "path": "poster_ocr_1024/6da895a65f0a400491abf53a909785ce.png", "path_original": "poster_ocr/6da895a65f0a400491abf53a909785ce.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['&', 'SONI', 'DEV', '7. 11. 2015', 'heroes -las vegas resorts - casino', 'flyerheroes.com presents', 'POOL SIDE HOUSE MUSIC']", "size": [ 1275, 1875 ], "texts": [ "&", "SONI", "DEV", "7. 11. 2015", "heroes -las vegas resorts - casino", "flyerheroes.com presents", "POOL SIDE HOUSE MUSIC" ], "text_bbox": [ [ 732, 621, 1083, 1033 ], [ 251, 646, 748, 814 ], [ 290, 834, 743, 1013 ], [ 399, 1473, 900, 1562 ], [ 214, 1597, 1060, 1628 ], [ 320, 584, 961, 607 ], [ 371, 1051, 910, 1074 ] ], "bbox_areas": [ 144612, 83496, 81087, 44589, 26226, 14743, 12397 ] }, { "task": "poster detection", "subtask": "", "name": "6d0f6c38d37643f887bcf217214bbb26.png", "path": "poster_ocr_1024/6d0f6c38d37643f887bcf217214bbb26.png", "path_original": "poster_ocr/6d0f6c38d37643f887bcf217214bbb26.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['york', 'new', 'dj angel / dj king', 'doors open: 9pm : free drinks : girls free till 10pm : only +21', 'contact: +123 456 7890 :: www.website.com', '25', 'satgur presents']", "size": [ 1275, 1875 ], "texts": [ "york", "new", "dj angel / dj king", "doors open: 9pm : free drinks : girls free till 10pm : only +21", "contact: +123 456 7890 :: www.website.com", "25", "satgur presents" ], "text_bbox": [ [ 377, 1316, 870, 1501 ], [ 417, 1116, 880, 1301 ], [ 100, 1611, 1182, 1653 ], [ 150, 1686, 1131, 1722 ], [ 283, 1738, 996, 1774 ], [ 566, 854, 726, 950 ], [ 418, 90, 899, 112 ] ], "bbox_areas": [ 91205, 85655, 45444, 35316, 25668, 15360, 10582 ] }, { "task": "poster detection", "subtask": "", "name": "6ccc22ef1d274b108e50f91ac7953385.png", "path": "poster_ocr_1024/6ccc22ef1d274b108e50f91ac7953385.png", "path_original": "poster_ocr/6ccc22ef1d274b108e50f91ac7953385.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['BUSINESSINSPIRATION DELICATIONS', 'To construcon project is compl Truc is complentcts a htyrt due to totrh them main tond otherentctsand otheron project is complentcts and otherentcts and other welcoem to maim.', '$ 99ONLY', 'Business Designs', 'Call us :+098 123 4556 234', 'Solutio-n', 'Prepaby julenjy touytum ']", "size": [ 2551, 3579 ], "texts": [ "BUSINESSINSPIRATION DELICATIONS", "To construcon project is compl Truc is complentcts a htyrt due to totrh them main tond otherentctsand otheron project is complentcts and otherentcts and other welcoem to maim.", "$ 99ONLY", "Business Designs", "Call us :+098 123 4556 234", "Solutio-n", "Prepaby julenjy touytum " ], "text_bbox": [ [ 220, 2804, 2032, 3088 ], [ 211, 3278, 2343, 3392 ], [ 990, 1367, 1552, 1780 ], [ 215, 3140, 1168, 3268 ], [ 1576, 242, 2328, 403 ], [ 351, 261, 1018, 354 ], [ 350, 372, 1008, 408 ] ], "bbox_areas": [ 514608, 243048, 232106, 121984, 121072, 62031, 23688 ] }, { "task": "poster detection", "subtask": "", "name": "6bf17c1dca63410292ff04a0d3eae71a.png", "path": "poster_ocr_1024/6bf17c1dca63410292ff04a0d3eae71a.png", "path_original": "poster_ocr/6bf17c1dca63410292ff04a0d3eae71a.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['CreativePortfolioProjects', 'Follow My Instagram', '#CreatoDesign', 'Vizit Website']", "size": [ 1200, 1200 ], "texts": [ "CreativePortfolioProjects", "Follow My Instagram", "#CreatoDesign", "Vizit Website" ], "text_bbox": [ [ 150, 437, 563, 748 ], [ 817, 93, 1050, 142 ], [ 240, 99, 486, 130 ], [ 212, 889, 337, 903 ] ], "bbox_areas": [ 128443, 11417, 7626, 1750 ] }, { "task": "poster detection", "subtask": "", "name": "6a61d938b86945dda125617ace79b5c5.png", "path": "poster_ocr_1024/6a61d938b86945dda125617ace79b5c5.png", "path_original": "poster_ocr/6a61d938b86945dda125617ace79b5c5.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['Makeuponline classes', 'Lorem ipsum dolor sit amet consectetur adipiscing elit, sed do.Lorem ipsum dolor sit amet consectetur adipiscing elit, sed do ', 'all daysposts & tutorials', 'www.makeup.com']", "size": [ 2480, 3508 ], "texts": [ "Makeuponline classes", "Lorem ipsum dolor sit amet consectetur adipiscing elit, sed do.Lorem ipsum dolor sit amet consectetur adipiscing elit, sed do ", "all daysposts & tutorials", "www.makeup.com" ], "text_bbox": [ [ 301, 597, 1486, 968 ], [ 304, 1089, 1514, 1295 ], [ 309, 1404, 816, 1534 ], [ 1964, 291, 2303, 327 ] ], "bbox_areas": [ 439635, 249260, 65910, 12204 ] }, { "task": "poster detection", "subtask": "", "name": "6a31ca8747b24fee91355d7b38971405.png", "path": "poster_ocr_1024/6a31ca8747b24fee91355d7b38971405.png", "path_original": "poster_ocr/6a31ca8747b24fee91355d7b38971405.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['GONE', 'GIRLZ', 'WILD', 'DJ MINIMAL . KAYNA YAZ . RON PAUL', '@WIXCLUB', 'www.yourwebsite.com', 'MIZUKO 77PRESENTS', '25NOV']", "size": [ 1275, 1875 ], "texts": [ "GONE", "GIRLZ", "WILD", "DJ MINIMAL . KAYNA YAZ . RON PAUL", "@WIXCLUB", "www.yourwebsite.com", "MIZUKO 77PRESENTS", "25NOV" ], "text_bbox": [ [ 292, 1156, 1005, 1358 ], [ 168, 965, 886, 1164 ], [ 302, 1348, 939, 1545 ], [ 141, 1715, 1140, 1752 ], [ 1018, 79, 1193, 242 ], [ 128, 1768, 1152, 1791 ], [ 80, 165, 300, 236 ], [ 942, 1432, 1062, 1543 ] ], "bbox_areas": [ 144026, 142882, 125489, 36963, 28525, 23552, 15620, 13320 ] }, { "task": "poster detection", "subtask": "", "name": "69ef73e74a27435e84fb366178422740.png", "path": "poster_ocr_1024/69ef73e74a27435e84fb366178422740.png", "path_original": "poster_ocr/69ef73e74a27435e84fb366178422740.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['SPACE', 'CITY', 'DJ JOHN / DJ RIVER / DJ ALEX', 'Doors open from 21PMat 123 Name Address, NY', 'PRESENT A NEW PARTY', 'www.creativemarket.com', 'PURCHASE TICKETS AT:', 'WE ARE HAPPY TO', 'WITH US WILL BE:', 'DEC.', '21']", "size": [ 1275, 1875 ], "texts": [ "SPACE", "CITY", "DJ JOHN / DJ RIVER / DJ ALEX", "Doors open from 21PMat 123 Name Address, NY", "PRESENT A NEW PARTY", "www.creativemarket.com", "PURCHASE TICKETS AT:", "WE ARE HAPPY TO", "WITH US WILL BE:", "DEC.", "21" ], "text_bbox": [ [ 434, 764, 840, 983 ], [ 525, 563, 754, 745 ], [ 270, 1405, 1008, 1450 ], [ 428, 1075, 848, 1143 ], [ 413, 125, 871, 156 ], [ 349, 1674, 928, 1696 ], [ 447, 1627, 865, 1656 ], [ 455, 76, 823, 107 ], [ 510, 1341, 766, 1363 ], [ 1042, 380, 1149, 419 ], [ 1068, 332, 1117, 370 ] ], "bbox_areas": [ 88914, 41678, 33210, 28560, 14198, 12738, 12122, 11408, 5632, 4173, 1862 ] }, { "task": "poster detection", "subtask": "", "name": "68ba516daf0240eaba5324764e9c5a65.png", "path": "poster_ocr_1024/68ba516daf0240eaba5324764e9c5a65.png", "path_original": "poster_ocr/68ba516daf0240eaba5324764e9c5a65.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['REVOLUTION', 'central city square, 10am, for more info please visit our webpage', 'stand up and fight for your rights', '24', 'Special place for aditional information, phone numbers, Names of members, facebook group links, etc.', 'www.revolutioninyourcity.com', 'of august', 'th']", "size": [ 2625, 3375 ], "texts": [ "REVOLUTION", "central city square, 10am, for more info please visit our webpage", "stand up and fight for your rights", "24", "Special place for aditional information, phone numbers, Names of members, facebook group links, etc.", "www.revolutioninyourcity.com", "of august", "th" ], "text_bbox": [ [ 308, 2288, 2352, 2733 ], [ 125, 3024, 2499, 3112 ], [ 481, 2805, 2147, 2909 ], [ 264, 285, 635, 690 ], [ 130, 3120, 2497, 3176 ], [ 796, 3192, 1759, 3259 ], [ 270, 707, 732, 811 ], [ 584, 293, 739, 442 ] ], "bbox_areas": [ 909580, 208912, 173264, 150255, 132552, 64521, 48048, 23095 ] }, { "task": "poster detection", "subtask": "", "name": "68906f79339c46e7bd9d03e0c862b80f.png", "path": "poster_ocr_1024/68906f79339c46e7bd9d03e0c862b80f.png", "path_original": "poster_ocr/68906f79339c46e7bd9d03e0c862b80f.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['26.06.2016', '26.06.2016', 'WATER', 'MUSIC', 'MUSIC', 'ANDROMEDE . DJ STTARS . GALAXY BLAST', 'ANDROMEDE . DJ STTARS . GALAXY BLAST', 'EDM . ELECTRO . DUBSTEP . HOUSE ', 'EDM . ELECTRO . DUBSTEP . HOUSE ', 'FIRE', 'www.yourclub.com', 'www.yourclub.com', 'EVENT', 'EVENT', 'BEST MINIMAL EVENT', 'BEST MINIMAL EVENT']", "size": [ 1275, 1875 ], "texts": [ "26.06.2016", "26.06.2016", "WATER", "MUSIC", "MUSIC", "ANDROMEDE . DJ STTARS . GALAXY BLAST", "ANDROMEDE . DJ STTARS . GALAXY BLAST", "EDM . ELECTRO . DUBSTEP . HOUSE ", "EDM . ELECTRO . DUBSTEP . HOUSE ", "FIRE", "www.yourclub.com", "www.yourclub.com", "EVENT", "EVENT", "BEST MINIMAL EVENT", "BEST MINIMAL EVENT" ], "text_bbox": [ [ 283, 1296, 999, 1375 ], [ 283, 1296, 999, 1375 ], [ 470, 574, 796, 677 ], [ 490, 699, 780, 803 ], [ 490, 699, 780, 803 ], [ 255, 1491, 1016, 1528 ], [ 255, 1491, 1016, 1528 ], [ 250, 1540, 1020, 1570 ], [ 250, 1540, 1020, 1570 ], [ 542, 574, 724, 677 ], [ 257, 1603, 997, 1627 ], [ 257, 1603, 997, 1627 ], [ 534, 824, 741, 898 ], [ 534, 824, 741, 898 ], [ 334, 1392, 932, 1417 ], [ 334, 1392, 932, 1417 ] ], "bbox_areas": [ 56564, 56564, 33578, 30160, 30160, 28157, 28157, 23100, 23100, 18746, 17760, 17760, 15318, 15318, 14950, 14950 ] }, { "task": "poster detection", "subtask": "", "name": "68213433dd024107b14155b29023d7bb.png", "path": "poster_ocr_1024/68213433dd024107b14155b29023d7bb.png", "path_original": "poster_ocr/68213433dd024107b14155b29023d7bb.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['ORIGINAL', 'The', '-FAT MINIMAL MUSIC PRESENTS-', 'DEEJAY MINIMAL . DEEJAY SOUND', 'ELECTRO / HOUSE / DUBSTEP', 'YOURCLUB', '19/04/2015', 'www.yourwebsite.com', 'EVENTS']", "size": [ 1275, 1875 ], "texts": [ "ORIGINAL", "The", "-FAT MINIMAL MUSIC PRESENTS-", "DEEJAY MINIMAL . DEEJAY SOUND", "ELECTRO / HOUSE / DUBSTEP", "YOURCLUB", "19/04/2015", "www.yourwebsite.com", "EVENTS" ], "text_bbox": [ [ 366, 858, 880, 1013 ], [ 539, 763, 729, 854 ], [ 413, 278, 858, 308 ], [ 410, 1470, 861, 1499 ], [ 425, 1508, 851, 1531 ], [ 539, 1575, 738, 1621 ], [ 532, 319, 747, 354 ], [ 511, 1629, 760, 1653 ], [ 542, 1037, 725, 1067 ] ], "bbox_areas": [ 79670, 17290, 13350, 13079, 9798, 9154, 7525, 5976, 5490 ] }, { "task": "poster detection", "subtask": "", "name": "67a605debd5441b3b074170b23f952b1.png", "path": "poster_ocr_1024/67a605debd5441b3b074170b23f952b1.png", "path_original": "poster_ocr/67a605debd5441b3b074170b23f952b1.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['Smoothie', 'RECIPES', '*New post on the blog.', 'YUMMMY!']", "size": [ 1080, 1920 ], "texts": [ "Smoothie", "RECIPES", "*New post on the blog.", "YUMMMY!" ], "text_bbox": [ [ 198, 159, 919, 522 ], [ 173, 501, 906, 597 ], [ 354, 1646, 725, 1675 ], [ 426, 1552, 652, 1578 ] ], "bbox_areas": [ 261723, 70368, 10759, 5876 ] }, { "task": "poster detection", "subtask": "", "name": "644435b1998a4c10b6924e2b0f812068.png", "path": "poster_ocr_1024/644435b1998a4c10b6924e2b0f812068.png", "path_original": "poster_ocr/644435b1998a4c10b6924e2b0f812068.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['INDIENIGHT', 'Daughters // the locust //holy molar // the ptd // Melt Banana // destroyed', '14 . 07 . 20168pm onwards', 'for more information go to www.yourclub.com or call +00123456789', 'G I R A F F F I k P R E S E N T S', '$20', 'entry']", "size": [ 2556, 3582 ], "texts": [ "INDIENIGHT", "Daughters // the locust //holy molar // the ptd // Melt Banana // destroyed", "14 . 07 . 20168pm onwards", "for more information go to www.yourclub.com or call +00123456789", "G I R A F F F I k P R E S E N T S", "$20", "entry" ], "text_bbox": [ [ 1636, 1143, 2387, 1861 ], [ 1435, 2779, 2194, 3025 ], [ 162, 3020, 864, 3252 ], [ 1434, 3217, 2152, 3438 ], [ 914, 236, 1606, 363 ], [ 331, 2272, 619, 2475 ], [ 337, 2508, 615, 2615 ] ], "bbox_areas": [ 539218, 186714, 162864, 158678, 87884, 58464, 29746 ] }, { "task": "poster detection", "subtask": "", "name": "630c3724e33c4fe3a1102b1295da2e6f.png", "path": "poster_ocr_1024/630c3724e33c4fe3a1102b1295da2e6f.png", "path_original": "poster_ocr/630c3724e33c4fe3a1102b1295da2e6f.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['BASS', 'BASS', 'BASS', 'ATTACK', 'DJ SLICK | DJ PAUL | MC DAVEY', '10$ DRINKS | FREE PARKING | DOORS OPEN AT 9PM', 'SATURDAY, MARCH 21ST', '7654 MAIN STREET, CITY CA. NEAR THE PLAZA WWW.WEBSITE.COM', 'ENTRY', 'CLUB ZEPPELIN', 'FREE', 'ALL NIGHT', 'PRESENTS', 'FACEBOOK', 'YOUTUBE', 'TWITTER', 'VIMEO']", "size": [ 2700, 3450 ], "texts": [ "BASS", "BASS", "BASS", "ATTACK", "DJ SLICK | DJ PAUL | MC DAVEY", "10$ DRINKS | FREE PARKING | DOORS OPEN AT 9PM", "SATURDAY, MARCH 21ST", "7654 MAIN STREET, CITY CA. NEAR THE PLAZA WWW.WEBSITE.COM", "ENTRY", "CLUB ZEPPELIN", "FREE", "ALL NIGHT", "PRESENTS", "FACEBOOK", "YOUTUBE", "TWITTER", "VIMEO" ], "text_bbox": [ [ 324, 1613, 2364, 2288 ], [ 342, 1620, 2346, 2284 ], [ 360, 1627, 2326, 2279 ], [ 515, 2203, 2186, 2779 ], [ 506, 2894, 2201, 3008 ], [ 376, 3018, 2312, 3094 ], [ 679, 2716, 2017, 2820 ], [ 502, 3158, 2198, 3203 ], [ 2005, 455, 2410, 586 ], [ 1012, 141, 1689, 212 ], [ 2075, 351, 2337, 466 ], [ 2041, 570, 2363, 644 ], [ 1178, 227, 1524, 285 ], [ 444, 3272, 712, 3310 ], [ 2097, 3270, 2334, 3308 ], [ 1004, 3274, 1221, 3310 ], [ 1570, 3272, 1735, 3310 ] ], "bbox_areas": [ 1377000, 1330656, 1281832, 962496, 193230, 147136, 139152, 76320, 53055, 48067, 30130, 23828, 20068, 10184, 9006, 7812, 6270 ] }, { "task": "poster detection", "subtask": "", "name": "62d306e360904e2b84a884bfe458cf35.png", "path": "poster_ocr_1024/62d306e360904e2b84a884bfe458cf35.png", "path_original": "poster_ocr/62d306e360904e2b84a884bfe458cf35.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['Back in season ', 'Double up on layers and texture to make a subtle,laidback, everyday look.', 'shop my favorite picks for the season', 'style tip', 'signature style']", "size": [ 750, 1050 ], "texts": [ "Back in season ", "Double up on layers and texture to make a subtle,laidback, everyday look.", "shop my favorite picks for the season", "style tip", "signature style" ], "text_bbox": [ [ 49, 160, 704, 218 ], [ 77, 624, 354, 732 ], [ 107, 249, 643, 286 ], [ 95, 554, 324, 594 ], [ 206, 100, 545, 124 ] ], "bbox_areas": [ 37990, 29916, 19832, 9160, 8136 ] }, { "task": "poster detection", "subtask": "", "name": "620c33207b5d4a4bb76d88701e76554e.png", "path": "poster_ocr_1024/620c33207b5d4a4bb76d88701e76554e.png", "path_original": "poster_ocr/620c33207b5d4a4bb76d88701e76554e.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['MUSICNIGHTS', '$99 Shot', 'Our mission is to controll and access all the ways thatcan build up success road', 'Our mission is to controll and access all the ways thatcan build up success road', 'Music Clubs', 'www.yourstudio.com', 'Pro#1', 'www.yourstudio.com', 'Prepared by Sulentum ']", "size": [ 3239, 4763 ], "texts": [ "MUSICNIGHTS", "$99 Shot", "Our mission is to controll and access all the ways thatcan build up success road", "Our mission is to controll and access all the ways thatcan build up success road", "Music Clubs", "www.yourstudio.com", "Pro#1", "www.yourstudio.com", "Prepared by Sulentum " ], "text_bbox": [ [ 744, 1886, 2535, 2698 ], [ 938, 2908, 2289, 3238 ], [ 998, 1600, 2280, 1784 ], [ 247, 278, 1545, 435 ], [ 2073, 4187, 2997, 4347 ], [ 1161, 2731, 2119, 2827 ], [ 1403, 1186, 1909, 1330 ], [ 245, 458, 1044, 536 ], [ 2388, 4386, 2986, 4443 ] ], "bbox_areas": [ 1454292, 445830, 235888, 203786, 147840, 91968, 72864, 62322, 34086 ] }, { "task": "poster detection", "subtask": "", "name": "61ef7be75830477da8d04ff1b904e436.png", "path": "poster_ocr_1024/61ef7be75830477da8d04ff1b904e436.png", "path_original": "poster_ocr/61ef7be75830477da8d04ff1b904e436.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['URBAN', 'SOUNDS', 'Friday 25th July', 'DOORS OPEN: 10PM | ENTRANCE: $20 + FREE DRINK | 21+ EVENT', 'www.urbansounds.com', 'DJ STEPHAN', 'DJ JACKSON', 'FOR MORE INFO VISIT', 'FEATURING ON THE DECKS', 'YOUR CLUB PRESENTS']", "size": [ 1575, 2175 ], "texts": [ "URBAN", "SOUNDS", "Friday 25th July", "DOORS OPEN: 10PM | ENTRANCE: $20 + FREE DRINK | 21+ EVENT", "www.urbansounds.com", "DJ STEPHAN", "DJ JACKSON", "FOR MORE INFO VISIT", "FEATURING ON THE DECKS", "YOUR CLUB PRESENTS" ], "text_bbox": [ [ 112, 1030, 1463, 1302 ], [ 286, 1250, 1288, 1422 ], [ 386, 1379, 1170, 1493 ], [ 158, 1836, 1447, 1901 ], [ 355, 1990, 1223, 2067 ], [ 157, 1699, 646, 1790 ], [ 955, 1699, 1440, 1790 ], [ 573, 1927, 1004, 1974 ], [ 549, 1505, 1027, 1546 ], [ 570, 71, 977, 112 ] ], "bbox_areas": [ 367472, 172344, 89376, 83785, 66836, 44499, 44135, 20257, 19598, 16687 ] }, { "task": "poster detection", "subtask": "", "name": "61b266bb80e14564ac688d391f27b4dc.png", "path": "poster_ocr_1024/61b266bb80e14564ac688d391f27b4dc.png", "path_original": "poster_ocr/61b266bb80e14564ac688d391f27b4dc.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['beautiful', 'Mamma Mia', 'Love is companionship']", "size": [ 3744, 5616 ], "texts": [ "beautiful", "Mamma Mia", "Love is companionship" ], "text_bbox": [ [ 1138, 1364, 2619, 1831 ], [ 712, 774, 2974, 1050 ], [ 902, 1121, 2884, 1240 ] ], "bbox_areas": [ 691627, 624312, 235858 ] }, { "task": "poster detection", "subtask": "", "name": "61995654bef549f59fdcafcb1769c47b.png", "path": "poster_ocr_1024/61995654bef549f59fdcafcb1769c47b.png", "path_original": "poster_ocr/61995654bef549f59fdcafcb1769c47b.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['COLORS', 'JUNUARY25-12-2014-8PMBEST PARTY EVER WORLD', 'BEST BUSINESS AGENCY FLYER TEMPLATEALL SOLUTIONS WITH US', 'Lorem Ipsum is simply dummy text of the printing and typesetting industry galley of type.', 'SPECAIL GUESTDJ ALEXA | DJ JOHN', 'ELECTRO', 'BEST FLYER WORLS EVER', 'PRESENTANNAAYA-220', '18.12.14']", "size": [ 1275, 1875 ], "texts": [ "COLORS", "JUNUARY25-12-2014-8PMBEST PARTY EVER WORLD", "BEST BUSINESS AGENCY FLYER TEMPLATEALL SOLUTIONS WITH US", "Lorem Ipsum is simply dummy text of the printing and typesetting industry galley of type.", "SPECAIL GUESTDJ ALEXA | DJ JOHN", "ELECTRO", "BEST FLYER WORLS EVER", "PRESENTANNAAYA-220", "18.12.14" ], "text_bbox": [ [ 373, 899, 910, 997 ], [ 76, 77, 507, 184 ], [ 294, 1725, 985, 1791 ], [ 75, 199, 619, 249 ], [ 470, 1108, 814, 1172 ], [ 461, 823, 816, 882 ], [ 357, 1669, 921, 1704 ], [ 514, 710, 769, 771 ], [ 510, 1019, 773, 1077 ] ], "bbox_areas": [ 52626, 46117, 45606, 27200, 22016, 20945, 19740, 15555, 15254 ] }, { "task": "poster detection", "subtask": "", "name": "60543c6affa94f18b25a3c361a63bc11.png", "path": "poster_ocr_1024/60543c6affa94f18b25a3c361a63bc11.png", "path_original": "poster_ocr/60543c6affa94f18b25a3c361a63bc11.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['LADIes NIGHT', 'EverySaturdays', 'dj LLOYD | Dj MONEY | DJ STRIKE', '9 PM || ENTRY 20$ || DRESS CODE : ALL IN BLACK', '-OVNI EVENT PRESENTs-', 'NightClub', 'FRIDAY 15th JULY', 'f facebook.com/YOURNAME', 'T twitter.com/YOURNAME']", "size": [ 1275, 1875 ], "texts": [ "LADIes NIGHT", "EverySaturdays", "dj LLOYD | Dj MONEY | DJ STRIKE", "9 PM || ENTRY 20$ || DRESS CODE : ALL IN BLACK", "-OVNI EVENT PRESENTs-", "NightClub", "FRIDAY 15th JULY", "f facebook.com/YOURNAME", "T twitter.com/YOURNAME" ], "text_bbox": [ [ 163, 1228, 1121, 1476 ], [ 310, 1125, 960, 1303 ], [ 241, 1536, 1034, 1588 ], [ 255, 1607, 1021, 1642 ], [ 277, 1082, 991, 1108 ], [ 525, 1658, 762, 1713 ], [ 491, 1026, 777, 1065 ], [ 254, 1658, 522, 1687 ], [ 772, 1658, 1019, 1687 ] ], "bbox_areas": [ 237584, 115700, 41236, 26810, 18564, 13035, 11154, 7772, 7163 ] }, { "task": "poster detection", "subtask": "", "name": "5c7bb5791f654411911f06b9cb22ed29.png", "path": "poster_ocr_1024/5c7bb5791f654411911f06b9cb22ed29.png", "path_original": "poster_ocr/5c7bb5791f654411911f06b9cb22ed29.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['SUMMER', 'MC JONES | DJ MARK | & SPECIAL GUESTS', 'BREAK', 'MAY 1TH', '1202 MAIN STREET, CITY NEW YORK, NEAR THE PLAZA', 'COCKTAILS AT THE BAR / FREE PARKING / DOORS OPEN AT 9PM', 'FACEBOOK/CLUBNAME TWITTER/CLUBNAME YOUTUBE/CLUBNAME', '@ ENVATO NIGHT CLUB', '25$', 'A GRAPHIC PRESENTS', 'ENTRY', 'WWW.WEBSITE.COM', '5566 664 443']", "size": [ 2700, 3450 ], "texts": [ "SUMMER", "MC JONES | DJ MARK | & SPECIAL GUESTS", "BREAK", "MAY 1TH", "1202 MAIN STREET, CITY NEW YORK, NEAR THE PLAZA", "COCKTAILS AT THE BAR / FREE PARKING / DOORS OPEN AT 9PM", "FACEBOOK/CLUBNAME TWITTER/CLUBNAME YOUTUBE/CLUBNAME", "@ ENVATO NIGHT CLUB", "25$", "A GRAPHIC PRESENTS", "ENTRY", "WWW.WEBSITE.COM", "5566 664 443" ], "text_bbox": [ [ 661, 1531, 2041, 1769 ], [ 576, 429, 2122, 511 ], [ 1012, 1801, 1689, 1952 ], [ 1010, 2126, 1687, 2260 ], [ 576, 2756, 2120, 2814 ], [ 576, 2699, 2126, 2745 ], [ 578, 3143, 2127, 3184 ], [ 1012, 1994, 1679, 2059 ], [ 625, 624, 894, 751 ], [ 1086, 254, 1615, 301 ], [ 623, 752, 895, 813 ], [ 1727, 2871, 2125, 2906 ], [ 1856, 2914, 2121, 2951 ] ], "bbox_areas": [ 328440, 126772, 102227, 90718, 89552, 71300, 63509, 43355, 34163, 24863, 16592, 13930, 9805 ] }, { "task": "poster detection", "subtask": "", "name": "5b307fdaf0454e09b9617fa9691d4bd6.png", "path": "poster_ocr_1024/5b307fdaf0454e09b9617fa9691d4bd6.png", "path_original": "poster_ocr/5b307fdaf0454e09b9617fa9691d4bd6.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['sunset', 'sunset', '10.30am - 1.30am', '10.30am - 1.30am', 'freeappetizer', 'exquisite tapas with exotic cocktails. Ibiza', 'exquisite tapas with exotic cocktails. Ibiza', 'with all bottles ofwine ordered', 'cafe del...', 'cafe del...', 'Food served all night. Doors open:']", "size": [ 2551, 3579 ], "texts": [ "sunset", "sunset", "10.30am - 1.30am", "10.30am - 1.30am", "freeappetizer", "exquisite tapas with exotic cocktails. Ibiza", "exquisite tapas with exotic cocktails. Ibiza", "with all bottles ofwine ordered", "cafe del...", "cafe del...", "Food served all night. Doors open:" ], "text_bbox": [ [ 286, 2103, 2103, 2735 ], [ 296, 2103, 2113, 2735 ], [ 297, 3052, 1566, 3297 ], [ 305, 3052, 1574, 3297 ], [ 297, 1021, 993, 1425 ], [ 297, 2783, 2055, 2907 ], [ 303, 2783, 2061, 2907 ], [ 345, 1431, 994, 1663 ], [ 295, 1884, 992, 2053 ], [ 305, 1884, 1002, 2053 ], [ 295, 3021, 992, 3069 ] ], "bbox_areas": [ 1148344, 1148344, 310905, 310905, 281184, 217992, 217992, 150568, 117793, 117793, 33456 ] }, { "task": "poster detection", "subtask": "", "name": "55152f97f7eb4e6996cca2d29a609a3c.png", "path": "poster_ocr_1024/55152f97f7eb4e6996cca2d29a609a3c.png", "path_original": "poster_ocr/55152f97f7eb4e6996cca2d29a609a3c.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['World forest day', 'Lorem ipsum dolor sit amet, consectetur adipiscing elit sed do eiusmod tempor incididunt ut labore et dolore.', 'protect our Forest ', '21 March', 'your account', 'your account', 'your account', 'www.yoursite.com']", "size": [ 2480, 3508 ], "texts": [ "World forest day", "Lorem ipsum dolor sit amet, consectetur adipiscing elit sed do eiusmod tempor incididunt ut labore et dolore.", "protect our Forest ", "21 March", "your account", "your account", "your account", "www.yoursite.com" ], "text_bbox": [ [ 569, 2114, 2007, 2625 ], [ 196, 161, 1325, 388 ], [ 650, 2796, 1908, 2886 ], [ 1038, 2629, 1500, 2710 ], [ 1124, 3279, 1526, 3338 ], [ 441, 3284, 843, 3343 ], [ 1795, 3273, 2197, 3332 ], [ 242, 526, 836, 563 ] ], "bbox_areas": [ 734818, 256283, 113220, 37422, 23718, 23718, 23718, 21978 ] }, { "task": "poster detection", "subtask": "", "name": "50439a1e9f874e9c9c74d4033bce3668.png", "path": "poster_ocr_1024/50439a1e9f874e9c9c74d4033bce3668.png", "path_original": "poster_ocr/50439a1e9f874e9c9c74d4033bce3668.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['ELECTRONIGHT', 'your venueyour city, 41220', 'saturday24th july 20168pm onwards', 'for more information go to www.yourclub.com or call +00123456789', 'G I R A F F F I k P R E S E N T S', 'DJ MIDNIGHT', 'DJ RUNNERS', 'FEATURING']", "size": [ 2556, 3582 ], "texts": [ "ELECTRONIGHT", "your venueyour city, 41220", "saturday24th july 20168pm onwards", "for more information go to www.yourclub.com or call +00123456789", "G I R A F F F I k P R E S E N T S", "DJ MIDNIGHT", "DJ RUNNERS", "FEATURING" ], "text_bbox": [ [ 733, 1467, 1823, 1927 ], [ 957, 2116, 1612, 2347 ], [ 1075, 1030, 1493, 1291 ], [ 839, 3271, 1727, 3387 ], [ 931, 201, 1601, 324 ], [ 478, 2816, 1064, 2923 ], [ 1531, 2816, 2092, 2923 ], [ 1149, 2614, 1418, 2677 ] ], "bbox_areas": [ 501400, 151305, 109098, 103008, 82410, 62702, 60027, 16947 ] }, { "task": "poster detection", "subtask": "", "name": "4dfff34bd4774eb5bb6632742c56fc65.png", "path": "poster_ocr_1024/4dfff34bd4774eb5bb6632742c56fc65.png", "path_original": "poster_ocr/4dfff34bd4774eb5bb6632742c56fc65.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['SPACEDREAM', 'FEATURINGDEEJAY ENVATO // DEEJAY GRAPHICS', 'HOSTED BY ENVATO', 'ELECTRO / HOUSE / DUBSTEP/ECLECTIC', '15TH AUGUST 2014', 'ENTRY 10$ free admission for ladies until 10', 'YOURNAME PRESENTS', 'TICKET INFO AND LATEST NEWS ON WWW.WEBSITE.COM', 'NightClub', 'www.facebook.com', 'www.twitter.com', 'www.yourwebsite.com']", "size": [ 1311, 1819 ], "texts": [ "SPACEDREAM", "FEATURINGDEEJAY ENVATO // DEEJAY GRAPHICS", "HOSTED BY ENVATO", "ELECTRO / HOUSE / DUBSTEP/ECLECTIC", "15TH AUGUST 2014", "ENTRY 10$ free admission for ladies until 10", "YOURNAME PRESENTS", "TICKET INFO AND LATEST NEWS ON WWW.WEBSITE.COM", "NightClub", "www.facebook.com", "www.twitter.com", "www.yourwebsite.com" ], "text_bbox": [ [ 265, 345, 1028, 849 ], [ 248, 1139, 1065, 1234 ], [ 291, 904, 1012, 958 ], [ 248, 1236, 1066, 1279 ], [ 335, 1032, 983, 1079 ], [ 248, 1335, 1066, 1370 ], [ 341, 211, 979, 243 ], [ 247, 1376, 1066, 1399 ], [ 551, 1580, 773, 1632 ], [ 363, 1470, 626, 1493 ], [ 771, 1473, 990, 1495 ], [ 552, 1639, 772, 1660 ] ], "bbox_areas": [ 384552, 77615, 38934, 35174, 30456, 28630, 20416, 18837, 11544, 6049, 4818, 4620 ] }, { "task": "poster detection", "subtask": "", "name": "4acb3d91f29446a78b031efe86fd29a5.png", "path": "poster_ocr_1024/4acb3d91f29446a78b031efe86fd29a5.png", "path_original": "poster_ocr/4acb3d91f29446a78b031efe86fd29a5.png", "prompt": "You are a vision-language model assistant for text detection. Given an image and a list of text elements, return a Python list of normalized bounding boxes in the format [[xmin, ymin, xmax, ymax], ...]. Each coordinate should be: \n 1. Expressed as decimals relative to the image's width (x-axis) and height (y-axis) \n 2. Precise to exactly 3 decimal places \n 3. Ordered as [left, top, right, bottom] in normalized coordinates. \n eg. [[0.123, 0.456, 0.789, 0.901],[0.050, 0.112, 0.950, 0.188],[0.001, 0.923, 0.999, 0.987]] \nReturn only the list (empty if no matches). No explanations. Text Elements to locate: \n['Wanderlust', 'Lorem ipsum dolor sit amet, has fugit euripidis at. Id tacimatessensibus Per, vix falli sapientem ne. Ad sea meliuspercipitur interpretaris.', 'thefavorite places& travel tips.']", "size": [ 1000, 1000 ], "texts": [ "Wanderlust", "Lorem ipsum dolor sit amet, has fugit euripidis at. Id tacimatessensibus Per, vix falli sapientem ne. Ad sea meliuspercipitur interpretaris.", "thefavorite places& travel tips." ], "text_bbox": [ [ 59, 43, 1000, 492 ], [ 67, 852, 766, 937 ], [ 69, 701, 479, 815 ] ], "bbox_areas": [ 422509, 59415, 46740 ] } ]