Hugging Face's logo

Datasets:
hexuan21
/
VideoScore_legacy

License:
Dataset card Data Studio Files
xet
Community
hexuan21 commited on
Commit
44fb77b
·
verified ·
1 Parent(s): 5cece53

upload refined_40k/test_bench 's json, videos and frames

Browse files
Files changed (1) hide show
  1. data/refined_40k/test_bench/data_bad_mi2v.json +312 -0
data/refined_40k/test_bench/data_bad_mi2v.json ADDED
@@ -0,0 +1,312 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ [
2
+ {
3
+ "id": "f000569",
4
+ "images": [
5
+ "f000569_00.jpg",
6
+ "f000569_01.jpg",
7
+ "f000569_02.jpg",
8
+ "f000569_03.jpg",
9
+ "f000569_04.jpg",
10
+ "f000569_05.jpg",
11
+ "f000569_06.jpg",
12
+ "f000569_07.jpg",
13
+ "f000569_08.jpg",
14
+ "f000569_09.jpg",
15
+ "f000569_10.jpg",
16
+ "f000569_11.jpg",
17
+ "f000569_12.jpg",
18
+ "f000569_13.jpg",
19
+ "f000569_14.jpg",
20
+ "f000569_15.jpg"
21
+ ],
22
+ "conversations": [
23
+ {
24
+ "from": "human",
25
+ "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos, \nplease watch the following frames of a given video and see the text prompt for generating the video, \nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, the consistency of objects or humans in video\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nFor each dimension, output a number from [1,2,3,4], \nin which '1' means 'Bad', '2' means 'Average', '3' means 'Good', \n'4' means 'Real' or 'Perfect' (the video is like a real video)\nHere is an output example:\nvisual quality: 4\ntemporal consistency: 4\ndynamic degree: 3\ntext-to-video alignment: 1\nfactual consistency: 2\n\nFor this video, the text prompt is \"There is a room with colorful walls and shelves filled with various items such as glass jars, vases, bottles, and cups. there are also several paintings on the wall, including one of a woman in a blue dress and another of a man wearing a hat.\",\nall the frames of video are as follows: \n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> \n"
26
+ },
27
+ {
28
+ "from": "gpt",
29
+ "value": "visual quality: 2\n temporal consistency: 1\n dynamic degree: 3\n text-to-video alignment: 1\n factual consistency: 1\n"
30
+ }
31
+ ]
32
+ },
33
+ {
34
+ "id": "f000855",
35
+ "images": [
36
+ "f000855_00.jpg",
37
+ "f000855_01.jpg",
38
+ "f000855_02.jpg",
39
+ "f000855_03.jpg",
40
+ "f000855_04.jpg",
41
+ "f000855_05.jpg",
42
+ "f000855_06.jpg",
43
+ "f000855_07.jpg",
44
+ "f000855_08.jpg",
45
+ "f000855_09.jpg",
46
+ "f000855_10.jpg",
47
+ "f000855_11.jpg",
48
+ "f000855_12.jpg",
49
+ "f000855_13.jpg",
50
+ "f000855_14.jpg",
51
+ "f000855_15.jpg"
52
+ ],
53
+ "conversations": [
54
+ {
55
+ "from": "human",
56
+ "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos, \nplease watch the following frames of a given video and see the text prompt for generating the video, \nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, the consistency of objects or humans in video\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nFor each dimension, output a number from [1,2,3,4], \nin which '1' means 'Bad', '2' means 'Average', '3' means 'Good', \n'4' means 'Real' or 'Perfect' (the video is like a real video)\nHere is an output example:\nvisual quality: 4\ntemporal consistency: 4\ndynamic degree: 3\ntext-to-video alignment: 1\nfactual consistency: 2\n\nFor this video, the text prompt is \"In the ancient Islamic era, the Caliph is asleep in his luxurious bed in the palace as he reminisces. \",\nall the frames of video are as follows: \n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> \n"
57
+ },
58
+ {
59
+ "from": "gpt",
60
+ "value": "visual quality: 2\n temporal consistency: 1\n dynamic degree: 3\n text-to-video alignment: 1\n factual consistency: 1\n"
61
+ }
62
+ ]
63
+ },
64
+ {
65
+ "id": "f000637",
66
+ "images": [
67
+ "f000637_00.jpg",
68
+ "f000637_01.jpg",
69
+ "f000637_02.jpg",
70
+ "f000637_03.jpg",
71
+ "f000637_04.jpg",
72
+ "f000637_05.jpg",
73
+ "f000637_06.jpg",
74
+ "f000637_07.jpg",
75
+ "f000637_08.jpg",
76
+ "f000637_09.jpg",
77
+ "f000637_10.jpg",
78
+ "f000637_11.jpg",
79
+ "f000637_12.jpg",
80
+ "f000637_13.jpg",
81
+ "f000637_14.jpg",
82
+ "f000637_15.jpg"
83
+ ],
84
+ "conversations": [
85
+ {
86
+ "from": "human",
87
+ "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos, \nplease watch the following frames of a given video and see the text prompt for generating the video, \nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, the consistency of objects or humans in video\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nFor each dimension, output a number from [1,2,3,4], \nin which '1' means 'Bad', '2' means 'Average', '3' means 'Good', \n'4' means 'Real' or 'Perfect' (the video is like a real video)\nHere is an output example:\nvisual quality: 4\ntemporal consistency: 4\ndynamic degree: 3\ntext-to-video alignment: 1\nfactual consistency: 2\n\nFor this video, the text prompt is \"a team meeting in a futurestic office \",\nall the frames of video are as follows: \n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> \n"
88
+ },
89
+ {
90
+ "from": "gpt",
91
+ "value": "visual quality: 2\n temporal consistency: 1\n dynamic degree: 3\n text-to-video alignment: 1\n factual consistency: 1\n"
92
+ }
93
+ ]
94
+ },
95
+ {
96
+ "id": "f000127",
97
+ "images": [
98
+ "f000127_00.jpg",
99
+ "f000127_01.jpg",
100
+ "f000127_02.jpg",
101
+ "f000127_03.jpg",
102
+ "f000127_04.jpg",
103
+ "f000127_05.jpg",
104
+ "f000127_06.jpg",
105
+ "f000127_07.jpg",
106
+ "f000127_08.jpg",
107
+ "f000127_09.jpg",
108
+ "f000127_10.jpg",
109
+ "f000127_11.jpg",
110
+ "f000127_12.jpg",
111
+ "f000127_13.jpg",
112
+ "f000127_14.jpg",
113
+ "f000127_15.jpg"
114
+ ],
115
+ "conversations": [
116
+ {
117
+ "from": "human",
118
+ "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos, \nplease watch the following frames of a given video and see the text prompt for generating the video, \nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, the consistency of objects or humans in video\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nFor each dimension, output a number from [1,2,3,4], \nin which '1' means 'Bad', '2' means 'Average', '3' means 'Good', \n'4' means 'Real' or 'Perfect' (the video is like a real video)\nHere is an output example:\nvisual quality: 4\ntemporal consistency: 4\ndynamic degree: 3\ntext-to-video alignment: 1\nfactual consistency: 2\n\nFor this video, the text prompt is \"An astronaut and an alien exchanging gifts on a distant planet, under multiple moons, 3D, Pixar style, speed 6x, 120fps \",\nall the frames of video are as follows: \n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> \n"
119
+ },
120
+ {
121
+ "from": "gpt",
122
+ "value": "visual quality: 2\n temporal consistency: 1\n dynamic degree: 3\n text-to-video alignment: 1\n factual consistency: 1\n"
123
+ }
124
+ ]
125
+ },
126
+ {
127
+ "id": "f000080",
128
+ "images": [
129
+ "f000080_00.jpg",
130
+ "f000080_01.jpg",
131
+ "f000080_02.jpg",
132
+ "f000080_03.jpg",
133
+ "f000080_04.jpg",
134
+ "f000080_05.jpg",
135
+ "f000080_06.jpg",
136
+ "f000080_07.jpg",
137
+ "f000080_08.jpg",
138
+ "f000080_09.jpg",
139
+ "f000080_10.jpg",
140
+ "f000080_11.jpg",
141
+ "f000080_12.jpg",
142
+ "f000080_13.jpg",
143
+ "f000080_14.jpg",
144
+ "f000080_15.jpg"
145
+ ],
146
+ "conversations": [
147
+ {
148
+ "from": "human",
149
+ "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos, \nplease watch the following frames of a given video and see the text prompt for generating the video, \nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, the consistency of objects or humans in video\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nFor each dimension, output a number from [1,2,3,4], \nin which '1' means 'Bad', '2' means 'Average', '3' means 'Good', \n'4' means 'Real' or 'Perfect' (the video is like a real video)\nHere is an output example:\nvisual quality: 4\ntemporal consistency: 4\ndynamic degree: 3\ntext-to-video alignment: 1\nfactual consistency: 2\n\nFor this video, the text prompt is \"A dark room in minecraft.\",\nall the frames of video are as follows: \n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> \n"
150
+ },
151
+ {
152
+ "from": "gpt",
153
+ "value": "visual quality: 2\n temporal consistency: 1\n dynamic degree: 3\n text-to-video alignment: 1\n factual consistency: 1\n"
154
+ }
155
+ ]
156
+ },
157
+ {
158
+ "id": "f000950",
159
+ "images": [
160
+ "f000950_00.jpg",
161
+ "f000950_01.jpg",
162
+ "f000950_02.jpg",
163
+ "f000950_03.jpg",
164
+ "f000950_04.jpg",
165
+ "f000950_05.jpg",
166
+ "f000950_06.jpg",
167
+ "f000950_07.jpg",
168
+ "f000950_08.jpg",
169
+ "f000950_09.jpg",
170
+ "f000950_10.jpg",
171
+ "f000950_11.jpg",
172
+ "f000950_12.jpg",
173
+ "f000950_13.jpg",
174
+ "f000950_14.jpg",
175
+ "f000950_15.jpg"
176
+ ],
177
+ "conversations": [
178
+ {
179
+ "from": "human",
180
+ "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos, \nplease watch the following frames of a given video and see the text prompt for generating the video, \nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, the consistency of objects or humans in video\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nFor each dimension, output a number from [1,2,3,4], \nin which '1' means 'Bad', '2' means 'Average', '3' means 'Good', \n'4' means 'Real' or 'Perfect' (the video is like a real video)\nHere is an output example:\nvisual quality: 4\ntemporal consistency: 4\ndynamic degree: 3\ntext-to-video alignment: 1\nfactual consistency: 2\n\nFor this video, the text prompt is \"A guy is seen near the boats.\n\",\nall the frames of video are as follows: \n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> \n"
181
+ },
182
+ {
183
+ "from": "gpt",
184
+ "value": "visual quality: 2\n temporal consistency: 1\n dynamic degree: 3\n text-to-video alignment: 1\n factual consistency: 1\n"
185
+ }
186
+ ]
187
+ },
188
+ {
189
+ "id": "f000390",
190
+ "images": [
191
+ "f000390_00.jpg",
192
+ "f000390_01.jpg",
193
+ "f000390_02.jpg",
194
+ "f000390_03.jpg",
195
+ "f000390_04.jpg",
196
+ "f000390_05.jpg",
197
+ "f000390_06.jpg",
198
+ "f000390_07.jpg",
199
+ "f000390_08.jpg",
200
+ "f000390_09.jpg",
201
+ "f000390_10.jpg",
202
+ "f000390_11.jpg",
203
+ "f000390_12.jpg",
204
+ "f000390_13.jpg",
205
+ "f000390_14.jpg",
206
+ "f000390_15.jpg"
207
+ ],
208
+ "conversations": [
209
+ {
210
+ "from": "human",
211
+ "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos, \nplease watch the following frames of a given video and see the text prompt for generating the video, \nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, the consistency of objects or humans in video\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nFor each dimension, output a number from [1,2,3,4], \nin which '1' means 'Bad', '2' means 'Average', '3' means 'Good', \n'4' means 'Real' or 'Perfect' (the video is like a real video)\nHere is an output example:\nvisual quality: 4\ntemporal consistency: 4\ndynamic degree: 3\ntext-to-video alignment: 1\nfactual consistency: 2\n\nFor this video, the text prompt is \"a Bangladeshi student going to study abroad \",\nall the frames of video are as follows: \n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> \n"
212
+ },
213
+ {
214
+ "from": "gpt",
215
+ "value": "visual quality: 2\n temporal consistency: 1\n dynamic degree: 3\n text-to-video alignment: 1\n factual consistency: 1\n"
216
+ }
217
+ ]
218
+ },
219
+ {
220
+ "id": "f000648",
221
+ "images": [
222
+ "f000648_00.jpg",
223
+ "f000648_01.jpg",
224
+ "f000648_02.jpg",
225
+ "f000648_03.jpg",
226
+ "f000648_04.jpg",
227
+ "f000648_05.jpg",
228
+ "f000648_06.jpg",
229
+ "f000648_07.jpg",
230
+ "f000648_08.jpg",
231
+ "f000648_09.jpg",
232
+ "f000648_10.jpg",
233
+ "f000648_11.jpg",
234
+ "f000648_12.jpg",
235
+ "f000648_13.jpg",
236
+ "f000648_14.jpg",
237
+ "f000648_15.jpg"
238
+ ],
239
+ "conversations": [
240
+ {
241
+ "from": "human",
242
+ "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos, \nplease watch the following frames of a given video and see the text prompt for generating the video, \nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, the consistency of objects or humans in video\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nFor each dimension, output a number from [1,2,3,4], \nin which '1' means 'Bad', '2' means 'Average', '3' means 'Good', \n'4' means 'Real' or 'Perfect' (the video is like a real video)\nHere is an output example:\nvisual quality: 4\ntemporal consistency: 4\ndynamic degree: 3\ntext-to-video alignment: 1\nfactual consistency: 2\n\nFor this video, the text prompt is \"camera pans in on red roof top\",\nall the frames of video are as follows: \n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> \n"
243
+ },
244
+ {
245
+ "from": "gpt",
246
+ "value": "visual quality: 2\n temporal consistency: 1\n dynamic degree: 3\n text-to-video alignment: 1\n factual consistency: 1\n"
247
+ }
248
+ ]
249
+ },
250
+ {
251
+ "id": "f000300",
252
+ "images": [
253
+ "f000300_00.jpg",
254
+ "f000300_01.jpg",
255
+ "f000300_02.jpg",
256
+ "f000300_03.jpg",
257
+ "f000300_04.jpg",
258
+ "f000300_05.jpg",
259
+ "f000300_06.jpg",
260
+ "f000300_07.jpg",
261
+ "f000300_08.jpg",
262
+ "f000300_09.jpg",
263
+ "f000300_10.jpg",
264
+ "f000300_11.jpg",
265
+ "f000300_12.jpg",
266
+ "f000300_13.jpg",
267
+ "f000300_14.jpg",
268
+ "f000300_15.jpg"
269
+ ],
270
+ "conversations": [
271
+ {
272
+ "from": "human",
273
+ "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos, \nplease watch the following frames of a given video and see the text prompt for generating the video, \nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, the consistency of objects or humans in video\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nFor each dimension, output a number from [1,2,3,4], \nin which '1' means 'Bad', '2' means 'Average', '3' means 'Good', \n'4' means 'Real' or 'Perfect' (the video is like a real video)\nHere is an output example:\nvisual quality: 4\ntemporal consistency: 4\ndynamic degree: 3\ntext-to-video alignment: 1\nfactual consistency: 2\n\nFor this video, the text prompt is \"A black suv on display at an auto show.\",\nall the frames of video are as follows: \n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> \n"
274
+ },
275
+ {
276
+ "from": "gpt",
277
+ "value": "visual quality: 2\n temporal consistency: 1\n dynamic degree: 3\n text-to-video alignment: 1\n factual consistency: 1\n"
278
+ }
279
+ ]
280
+ },
281
+ {
282
+ "id": "f000553",
283
+ "images": [
284
+ "f000553_00.jpg",
285
+ "f000553_01.jpg",
286
+ "f000553_02.jpg",
287
+ "f000553_03.jpg",
288
+ "f000553_04.jpg",
289
+ "f000553_05.jpg",
290
+ "f000553_06.jpg",
291
+ "f000553_07.jpg",
292
+ "f000553_08.jpg",
293
+ "f000553_09.jpg",
294
+ "f000553_10.jpg",
295
+ "f000553_11.jpg",
296
+ "f000553_12.jpg",
297
+ "f000553_13.jpg",
298
+ "f000553_14.jpg",
299
+ "f000553_15.jpg"
300
+ ],
301
+ "conversations": [
302
+ {
303
+ "from": "human",
304
+ "value": "\nSuppose you are an expert in judging and evaluating the quality of AI-generated videos, \nplease watch the following frames of a given video and see the text prompt for generating the video, \nthen give scores from 5 different dimensions:\n(1) visual quality: the quality of the video in terms of clearness, resolution, brightness, and color\n(2) temporal consistency, the consistency of objects or humans in video\n(3) dynamic degree, the degree of dynamic changes\n(4) text-to-video alignment, the alignment between the text prompt and the video content\n(5) factual consistency, the consistency of the video content with the common-sense and factual knowledge\n\nFor each dimension, output a number from [1,2,3,4], \nin which '1' means 'Bad', '2' means 'Average', '3' means 'Good', \n'4' means 'Real' or 'Perfect' (the video is like a real video)\nHere is an output example:\nvisual quality: 4\ntemporal consistency: 4\ndynamic degree: 3\ntext-to-video alignment: 1\nfactual consistency: 2\n\nFor this video, the text prompt is \"Two people reaching out to touch a leopard in the wild.\",\nall the frames of video are as follows: \n\n<image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> <image> \n"
305
+ },
306
+ {
307
+ "from": "gpt",
308
+ "value": "visual quality: 2\n temporal consistency: 1\n dynamic degree: 3\n text-to-video alignment: 1\n factual consistency: 1\n"
309
+ }
310
+ ]
311
+ }
312
+ ]