Dataset Viewer
Auto-converted to Parquet Duplicate
id
stringlengths
18
24
patent_no
stringlengths
11
15
page_no
int64
1
312
pdf_path
stringlengths
35
41
preview_image
stringlengths
43
49
coordinate_reference_size
listlengths
2
2
layout_count
int64
1
52
ocsr_count
int64
0
32
layout_gt
listlengths
1
52
ocsr_gt
listlengths
0
32
page_image
imagewidth (px)
1.98k
2.12k
AU2022204980A1_page_156
AU2022204980A1
156
assets/pages/AU2022204980A1_page_156.pdf
assets/page_previews/AU2022204980A1_page_156.jpg
[ 4955, 7006 ]
4
4
[ { "bbox": [ 0.19182545862076034, 0.6640565077371532, 0.3580475761769389, 0.7517497946435179 ], "image": "assets/images/img-dc-5524.jpg", "labels": { "atom_bond_features": [], "chemical_family": [ "常规有机结构" ], "component_features": [], "ima...
[ { "inchikey": "ULGFAEQRKLBHCT-UHFFFAOYSA-N", "molfile": "assets/molfiles/img-dc-5524.mol", "region_id": "img-dc-5524", "smiles": "Cn1ccnc1-c1cc(Cl)cc(-c2nccn2C)n1" }, { "inchikey": "OEYPRQKLHCTTCV-UHFFFAOYSA-N", "molfile": "assets/molfiles/img-dc-5529.mol", "region_id": "img-dc-5529"...
CN107001339B_page_93
CN107001339B
93
assets/pages/CN107001339B_page_93.pdf
assets/page_previews/CN107001339B_page_93.jpg
[ 1190, 1684 ]
4
4
[ { "bbox": [ 0.18992133914317816, 0.0720131569583997, 0.3713153075601525, 0.11942301122935753 ], "image": "assets/images/img-dc-5341.jpg", "labels": { "atom_bond_features": [], "chemical_family": [ "常规有机结构" ], "component_features": [], "im...
[ { "inchikey": "KHXXGDIHQCGYPT-UKRRQHHQSA-N", "molfile": "assets/molfiles/img-dc-5341.mol", "region_id": "img-dc-5341", "smiles": "CC[C@@H]1CN(C#N)C[C@H]1C(=O)Nc1cn(-c2ccccc2)cn1" }, { "inchikey": "UACCRDNKUROTFE-UHFFFAOYSA-N", "molfile": "assets/molfiles/img-dc-5346.mol", "region_id"...
CN107918248B_page_20
CN107918248B
20
assets/pages/CN107918248B_page_20.pdf
assets/page_previews/CN107918248B_page_20.jpg
[ 1190, 1684 ]
32
32
[ { "bbox": [ 0.5200931633074305, 0.6365588804925673, 0.7007387077252843, 0.7126202910159057 ], "image": "assets/images/img-dc-6317.jpg", "labels": { "atom_bond_features": [ "显式电荷" ], "chemical_family": [ "常规有机结构" ], "component_feat...
[ { "inchikey": "GWDRSBRJDQBEFC-UHFFFAOYSA-N", "molfile": "assets/molfiles/img-dc-6317.mol", "region_id": "img-dc-6317", "smiles": "O=S(=O)([N-]c1cccc2ccccc12)C(F)(F)C(F)(F)C(F)(F)F" }, { "inchikey": "RUULAKUFARGBEV-UHFFFAOYSA-N", "molfile": "assets/molfiles/img-dc-6322.mol", "region_i...
CN108368425B_page_87
CN108368425B
87
assets/pages/CN108368425B_page_87.pdf
assets/page_previews/CN108368425B_page_87.jpg
[ 1190, 1684 ]
13
2
[ { "bbox": [ 0.18602333564395732, 0.3457167873719076, 0.34025382148524497, 0.396499307569694 ], "image": "assets/images/img-dc-1351.jpg", "labels": { "atom_bond_features": [], "chemical_family": [ "常规有机结构" ], "component_features": [], "ima...
[ { "inchikey": "NITWSHWHQAQBAW-QPJJXVBHSA-N", "molfile": "assets/molfiles/img-dc-1351.mol", "region_id": "img-dc-1351", "smiles": "COC(=O)/C=C/c1ccc(O)cc1" }, { "inchikey": "LXCFILQKKLGQFO-UHFFFAOYSA-N", "molfile": "assets/molfiles/img-dc-1341.mol", "region_id": "img-dc-1341", "sm...
CN109991810B_page_16
CN109991810B
16
assets/pages/CN109991810B_page_16.pdf
assets/page_previews/CN109991810B_page_16.jpg
[ 1190, 1684 ]
12
12
[ { "bbox": [ 0.20885194838047028, 0.4753433167934418, 0.40614354610443115, 0.5599316358566284 ], "image": "assets/images/img-dc-101.jpg", "labels": { "atom_bond_features": [], "chemical_family": [ "常规有机结构" ], "component_features": [], "ima...
[ { "inchikey": "HAZQZUFYRLFOLC-UHFFFAOYSA-N", "molfile": "assets/molfiles/img-dc-101.mol", "region_id": "img-dc-101", "smiles": "ClC(Cl)(Cl)c1nc(-c2ccccc2)nc(C(Cl)(Cl)Cl)n1" }, { "inchikey": "MCJPJAJHPRCILL-UHFFFAOYSA-N", "molfile": "assets/molfiles/img-dc-106.mol", "region_id": "img-...
CN110088253B_page_25
CN110088253B
25
assets/pages/CN110088253B_page_25.pdf
assets/page_previews/CN110088253B_page_25.jpg
[ 1190, 1684 ]
3
0
[ { "bbox": [ 0.2079539258595089, 0.15077009033627978, 0.4862838622300411, 0.22035291242289975 ], "image": "assets/images/img-dc-6794.jpg", "labels": { "atom_bond_features": [], "chemical_family": [ "常规有机结构" ], "component_features": [], "im...
[]
CN110997073B_page_17
CN110997073B
17
assets/pages/CN110997073B_page_17.pdf
assets/page_previews/CN110997073B_page_17.jpg
[ 1190, 1684 ]
21
0
[ { "bbox": [ 0.22555115876889253, 0.6904123815515811, 0.36204143623942464, 0.7486353359019773 ], "image": "assets/images/img-dc-7244.jpg", "labels": { "atom_bond_features": [], "chemical_family": [ "常规有机结构" ], "component_features": [], "im...
[]
CN111655726B_page_37
CN111655726B
37
assets/pages/CN111655726B_page_37.pdf
assets/page_previews/CN111655726B_page_37.jpg
[ 1191, 1684 ]
6
0
[ { "bbox": [ 0.29316906985548147, 0.48059988770873036, 0.6468492022617607, 0.5849396899104674 ], "image": "assets/images/img-dc-6102.jpg", "labels": { "atom_bond_features": [], "chemical_family": [ "生物偶联结构" ], "component_features": [], "im...
[]
CN112204009A_page_191
CN112204009A
191
assets/pages/CN112204009A_page_191.pdf
assets/page_previews/CN112204009A_page_191.jpg
[ 1190, 1684 ]
8
1
[ { "bbox": [ 0.18566028759892422, 0.4242287179148116, 0.2971747461301524, 0.47023632462221604 ], "image": "assets/images/img-dc-5432.jpg", "labels": { "atom_bond_features": [ "单独画出的氢" ], "chemical_family": [ "常规有机结构" ], "component_...
[ { "inchikey": "KMWLDIMOYZXPCV-UHFFFAOYSA-N", "molfile": "assets/molfiles/img-dc-5442.mol", "region_id": "img-dc-5442", "smiles": "Cc1nocc1C(=O)NC12CCC(NC(=O)COc3ccc(Cl)c(F)c3)(CC1)C(O)C2" } ]
CN112957463A_page_60
CN112957463A
60
assets/pages/CN112957463A_page_60.pdf
assets/page_previews/CN112957463A_page_60.jpg
[ 1190, 1684 ]
9
9
[ { "bbox": [ 0.581282377243042, 0.11540915071964264, 0.8687961101531982, 0.22060392796993256 ], "image": "assets/images/img-dc-217.jpg", "labels": { "atom_bond_features": [], "chemical_family": [ "肽/蛋白结构" ], "component_features": [], "imag...
[ { "inchikey": "KPHJDDKINWAPCS-VMPREFPWSA-N", "molfile": "assets/molfiles/img-dc-217.mol", "region_id": "img-dc-217", "smiles": "[H]N(C(=O)[C@H](CCCNC(N)=O)NC(=O)[C@@H](NC(=O)OCC1c2ccccc2-c2ccccc21)C(C)C)c1ccc(CN)cc1" }, { "inchikey": "OZLCQFUKDABFET-UHFFFAOYSA-N", "molfile": "assets/molf...
CN115298178A_page_27
CN115298178A
27
assets/pages/CN115298178A_page_27.pdf
assets/page_previews/CN115298178A_page_27.jpg
[ 1190, 1684 ]
8
8
[ { "bbox": [ 0.5736610889434814, 0.10450650751590729, 0.6922074556350708, 0.1610468259196937 ], "image": "assets/images/img-dc-5024.jpg", "labels": { "atom_bond_features": [], "chemical_family": [ "常规有机结构" ], "component_features": [], "ima...
[ { "inchikey": "HEKLOHIYSVDBCZ-UHFFFAOYSA-N", "molfile": "assets/molfiles/img-dc-5024.mol", "region_id": "img-dc-5024", "smiles": "CC(C)(C)OC(=O)N1CCCN(C(=O)OC(C)(C)C)CCN(C(=O)OC(C)(C)C)CCCN(CC(O)CBr)CC1" }, { "inchikey": "OUXNWTZVZBGUOH-UHFFFAOYSA-N", "molfile": "assets/molfiles/img-dc-5...
CN115515934A_page_312
CN115515934A
312
assets/pages/CN115515934A_page_312.pdf
assets/page_previews/CN115515934A_page_312.jpg
[ 1190, 1684 ]
3
3
[ { "bbox": [ 0.492610556145463, 0.081175533432629, 0.832412134904579, 0.16322695369662016 ], "image": "assets/images/img-dc-5822.jpg", "labels": { "atom_bond_features": [], "chemical_family": [ "常规有机结构" ], "component_features": [], "image_...
[ { "inchikey": "RXTMIEOMHPVZST-UHFFFAOYSA-N", "molfile": "assets/molfiles/img-dc-5822.mol", "region_id": "img-dc-5822", "smiles": "COc1ccc(CNC(=O)Nc2ccc(CNC(=O)C3CCOCC3)cc2)cc1" }, { "inchikey": "LCCYZXPDHMLVGS-UHFFFAOYSA-N", "molfile": "assets/molfiles/img-dc-5826.mol", "region_id": ...
CN115811976A_page_32
CN115811976A
32
assets/pages/CN115811976A_page_32.pdf
assets/page_previews/CN115811976A_page_32.jpg
[ 1190, 1684 ]
15
15
[ { "bbox": [ 0.4172709134388024, 0.0954782231963711, 0.5925255485302993, 0.2123760099369082 ], "image": "assets/images/img-dc-6077.jpg", "labels": { "atom_bond_features": [], "chemical_family": [ "常规有机结构" ], "component_features": [], "imag...
[ { "inchikey": "IYGPENJKPSNCOV-UHFFFAOYSA-N", "molfile": "assets/molfiles/img-dc-6077.mol", "region_id": "img-dc-6077", "smiles": "CC(=O)c1c(N)n(-c2c(C)ccc(O)c2C)c2ncc(-c3cnccn3)nc12" }, { "inchikey": "WQWJRZGKNFSCSW-UHFFFAOYSA-N", "molfile": "assets/molfiles/img-dc-6081.mol", "region...
CN119143833A_page_26
CN119143833A
26
assets/pages/CN119143833A_page_26.pdf
assets/page_previews/CN119143833A_page_26.jpg
[ 1191, 1684 ]
10
10
[ { "bbox": [ 0.3506980766504489, 0.267066085750458, 0.528934946221682, 0.3400922524282549 ], "image": "assets/images/img-dc-7157.jpg", "labels": { "atom_bond_features": [ "单独画出的氢" ], "chemical_family": [ "常规有机结构" ], "component_feat...
[ { "inchikey": "VUSCVWLUKLWUAW-RUCRIFKGSA-N", "molfile": "assets/molfiles/img-dc-7157.mol", "region_id": "img-dc-7157", "smiles": "[H][C@]12CC=C3[C@]4([H])C[C@@](C)(CNC(=O)NCC(C)C)CC[C@]4(C)CC[C@@]3(C)[C@]1(C)CC[C@@]1([H])C(C)(C)[C@@H](O)CC[C@]21C" }, { "inchikey": "OBXKKUJZXFAONU-OKJWXPAKSA-...
EP3904349B1_page_60
EP3904349B1
60
assets/pages/EP3904349B1_page_60.pdf
assets/page_previews/EP3904349B1_page_60.jpg
[ 1191, 1684 ]
8
8
[ { "bbox": [ 0.4707677591300808, 0.1390622366403237, 0.5380444873072223, 0.21497234542345495 ], "image": "assets/images/img-dc-6294.jpg", "labels": { "atom_bond_features": [], "chemical_family": [ "常规有机结构" ], "component_features": [], "ima...
[ { "inchikey": "GBTBMKLARFEFTP-UHFFFAOYSA-N", "molfile": "assets/molfiles/img-dc-6294.mol", "region_id": "img-dc-6294", "smiles": "N/C(=N\\Nc1ncccn1)C(F)(F)F" }, { "inchikey": "AMASJBTYNMAWDS-UHFFFAOYSA-N", "molfile": "assets/molfiles/img-dc-6300.mol", "region_id": "img-dc-6300", ...
HK40064761A_page_62
HK40064761A
62
assets/pages/HK40064761A_page_62.pdf
assets/page_previews/HK40064761A_page_62.jpg
[ 1190, 1684 ]
3
0
[ { "bbox": [ 0.1319694221019745, 0.14857534855951438, 0.8234992253239733, 0.3905541871984231 ], "image": "assets/images/img-dc-5093.jpg", "labels": { "atom_bond_features": [], "chemical_family": [ "肽/蛋白结构", "糖类结构", "聚合/重复结构", "生物偶联结构...
[]
JP2020105188A_page_162
JP2020105188A
162
assets/pages/JP2020105188A_page_162.pdf
assets/page_previews/JP2020105188A_page_162.jpg
[ 1191, 1684 ]
9
0
[ { "bbox": [ 0.5980346116943911, 0.2872980847562644, 0.7259358587838426, 0.3371964550643897 ], "image": "assets/images/img-dc-7343.jpg", "labels": { "atom_bond_features": [], "chemical_family": [ "常规有机结构" ], "component_features": [], "imag...
[]
JP2022527298A_page_263
JP2022527298A
263
assets/pages/JP2022527298A_page_263.pdf
assets/page_previews/JP2022527298A_page_263.jpg
[ 1190, 1684 ]
9
4
[ { "bbox": [ 0.29709617318432446, 0.07009109019123386, 0.5878848339980667, 0.15494648017373305 ], "image": "assets/images/img-dc-8141.jpg", "labels": { "atom_bond_features": [], "chemical_family": [ "肽/蛋白结构" ], "component_features": [], "i...
[ { "inchikey": "FBPKKRHWRMHIMM-LDVJMBRRSA-N", "molfile": "assets/molfiles/img-dc-8161.mol", "region_id": "img-dc-8161", "smiles": "CCOP(=O)(O)C(F)(F)c1ccc2sc(C(=O)N[C@H]3CN(C(=O)OC)CC[C@H]4CC[C@@H](C(=O)N[C@@H](CCC(N)=O)C(=O)OC(C)(C)C)N4C3=O)cc2c1" }, { "inchikey": "OLAUHIDAEKLXLP-FYQPLNBISA-...
JP7119204B2_page_47
JP7119204B2
47
assets/pages/JP7119204B2_page_47.pdf
assets/page_previews/JP7119204B2_page_47.jpg
[ 1190, 1684 ]
9
8
[ { "bbox": [ 0.3189709779066223, 0.7610966150860559, 0.46546494464095467, 0.8094328407893162 ], "image": "assets/images/img-dc-4525.jpg", "labels": { "atom_bond_features": [], "chemical_family": [ "常规有机结构" ], "component_features": [], "ima...
[ { "inchikey": "IPWKHHSGDUIRAH-UHFFFAOYSA-N", "molfile": "assets/molfiles/img-dc-4525.mol", "region_id": "img-dc-4525", "smiles": "CC1(C)OB(B2OC(C)(C)C(C)(C)O2)OC1(C)C" }, { "inchikey": "RMVRSNDYEFQCLF-UHFFFAOYSA-N", "molfile": "assets/molfiles/img-dc-4513.mol", "region_id": "img-dc-4...
JP7233871B2_page_39
JP7233871B2
39
assets/pages/JP7233871B2_page_39.pdf
assets/page_previews/JP7233871B2_page_39.jpg
[ 1190, 1684 ]
52
0
[ { "bbox": [ 0.422844668358182, 0.07431843396009348, 0.49777800604544015, 0.09763661614262188 ], "image": "assets/images/img-dc-4750.jpg", "labels": { "atom_bond_features": [], "chemical_family": [ "常规有机结构" ], "component_features": [], "im...
[]
JP7256130B2_page_14
JP7256130B2
14
assets/pages/JP7256130B2_page_14.pdf
assets/page_previews/JP7256130B2_page_14.jpg
[ 1190, 1684 ]
8
1
[ { "bbox": [ 0.10404212772846222, 0.07506781816482544, 0.2825235561488135, 0.18402436943174919 ], "image": "assets/images/img-dc-4792.jpg", "labels": { "atom_bond_features": [ "单独画出的氢" ], "chemical_family": [ "聚合/重复结构" ], "componen...
[ { "inchikey": "CFOILRQUSYGFMG-UHFFFAOYSA-N", "molfile": "assets/molfiles/img-dc-4824.mol", "region_id": "img-dc-4824", "smiles": "CC(C)(C)OC(=O)/N=C(/NC(=O)OC(C)(C)C)n1ccnc1" } ]
JP7478266B2_page_37
JP7478266B2
37
assets/pages/JP7478266B2_page_37.pdf
assets/page_previews/JP7478266B2_page_37.jpg
[ 1190, 1684 ]
12
12
[ { "bbox": [ 0.10521397494525533, 0.2786405970201448, 0.2541874828270689, 0.36405016093257697 ], "image": "assets/images/img-dc-2073.jpg", "labels": { "atom_bond_features": [], "chemical_family": [ "常规有机结构" ], "component_features": [], "im...
[ { "inchikey": "PTDAXLXIVYGHLC-UHFFFAOYSA-N", "molfile": "assets/molfiles/img-dc-2073.mol", "region_id": "img-dc-2073", "smiles": "c1ccc(-n2c3ccccc3c3c2ccc2c4ccccc4n(-c4cc5c6c(c4)-c4ccccc4OB6Oc4ccccc4-5)c23)cc1" }, { "inchikey": "VYSOVCXXEWBECK-UHFFFAOYSA-N", "molfile": "assets/molfiles/i...
KR102138351B1_page_38
KR102138351B1
38
assets/pages/KR102138351B1_page_38.pdf
assets/page_previews/KR102138351B1_page_38.jpg
[ 1191, 1684 ]
22
0
[ { "bbox": [ 0.4323926697943498, 0.35587194294336055, 0.567440817589163, 0.4112561073839405 ], "image": "assets/images/img-dc-6530.jpg", "labels": { "atom_bond_features": [], "chemical_family": [ "常规有机结构" ], "component_features": [], "imag...
[]
RU2024116271A_page_84
RU2024116271A
84
assets/pages/RU2024116271A_page_84.pdf
assets/page_previews/RU2024116271A_page_84.jpg
[ 1191, 1684 ]
13
0
[ { "bbox": [ 0.7103713750839233, 0.5292078256607056, 0.9017753601074219, 0.5839085578918457 ], "image": "assets/images/img-dc-4904.jpg", "labels": { "atom_bond_features": [], "chemical_family": [ "聚合/重复结构" ], "component_features": [], "ima...
[]
TNSN07202A1_page_7
TNSN07202A1
7
assets/pages/TNSN07202A1_page_7.pdf
assets/page_previews/TNSN07202A1_page_7.jpg
[ 1191, 1684 ]
2
0
[ { "bbox": [ 0.3651241000685577, 0.11368921871464115, 0.7029999934757276, 0.24906905010056102 ], "image": "assets/images/img-dc-6863.jpg", "labels": { "atom_bond_features": [ "查询/不确定键" ], "chemical_family": [ "聚合/重复结构" ], "componen...
[]
US10596270B2_page_103
US10596270B2
103
assets/pages/US10596270B2_page_103.pdf
assets/page_previews/US10596270B2_page_103.jpg
[ 1224, 1584 ]
4
4
[ { "bbox": [ 0.12447372624395901, 0.5469339959594782, 0.8323231542740859, 0.7197512806658417 ], "image": "assets/images/img-dc-7105.jpg", "labels": { "atom_bond_features": [], "chemical_family": [ "肽/蛋白结构", "生物偶联结构" ], "component_feature...
[ { "inchikey": "ZMSFFQZNTRZZJB-QUAMOIMISA-N", "molfile": "assets/molfiles/img-dc-7105.mol", "region_id": "img-dc-7105", "smiles": "CN[C@H](C(=O)N[C@H](C(=O)N(C)[C@H](/C=C(\\C)C(=O)O)C(C)C)C(C)(C)C)C(C)(C)c1cccc(NC(=O)OCc2ccc(NC(=O)[C@H](CCCNC(N)=O)NC(=O)[C@@H](NC(=O)CCC(=O)N3Cc4ccccc4C#Cc4ccccc43)C(C...
US10874686B2_page_10
US10874686B2
10
assets/pages/US10874686B2_page_10.pdf
assets/page_previews/US10874686B2_page_10.jpg
[ 1224, 1584 ]
2
0
[ { "bbox": [ 0.17367126049433115, 0.32256634727268796, 0.48656138653866654, 0.4543747537426844 ], "image": "assets/images/img-dc-3183.jpg", "labels": { "atom_bond_features": [ "显式电荷" ], "chemical_family": [ "核酸/核苷(酸)类结构", "糖类结构" ...
[]
US10961338B2_page_42
US10961338B2
42
assets/pages/US10961338B2_page_42.pdf
assets/page_previews/US10961338B2_page_42.jpg
[ 1224, 1584 ]
3
0
[ { "bbox": [ 0.13752534985542297, 0.5855005979537964, 0.4672757089138031, 0.7190076112747192 ], "image": "assets/images/img-dc-3265.jpg", "labels": { "atom_bond_features": [], "chemical_family": [ "聚合/重复结构" ], "component_features": [], "im...
[]
US11104649B2_page_116
US11104649B2
116
assets/pages/US11104649B2_page_116.pdf
assets/page_previews/US11104649B2_page_116.jpg
[ 1224, 1584 ]
9
9
[ { "bbox": [ 0.47251723498262305, 0.20539895554955923, 0.7183242695010105, 0.2817333191334793 ], "image": "assets/images/img-dc-7564.jpg", "labels": { "atom_bond_features": [], "chemical_family": [ "常规有机结构" ], "component_features": [], "im...
[ { "inchikey": "VOYSCQYOHPWPEV-UHFFFAOYSA-N", "molfile": "assets/molfiles/img-dc-7564.mol", "region_id": "img-dc-7564", "smiles": "O=C1c2ccccc2C(=O)N1CCCCCS(=O)(=O)Cl" }, { "inchikey": "NRTPGOWNOQCSEA-HNNXBMFYSA-N", "molfile": "assets/molfiles/img-dc-7581.mol", "region_id": "img-dc-75...
US11129820B2_page_96
US11129820B2
96
assets/pages/US11129820B2_page_96.pdf
assets/page_previews/US11129820B2_page_96.jpg
[ 1224, 1584 ]
28
27
[ { "bbox": [ 0.7079297257158703, 0.2345120770586228, 0.8432605281602017, 0.2944241510574153 ], "image": "assets/images/img-dc-3552.jpg", "labels": { "atom_bond_features": [ "显式电荷" ], "chemical_family": [ "常规有机结构" ], "component_feat...
[ { "inchikey": "BJEPYKJPYRNKOW-UHFFFAOYSA-M", "molfile": "assets/molfiles/img-dc-3552.mol", "region_id": "img-dc-3552", "smiles": "O=C([O-])CC(O)C(=O)O" }, { "inchikey": "BJEPYKJPYRNKOW-UHFFFAOYSA-L", "molfile": "assets/molfiles/img-dc-3556.mol", "region_id": "img-dc-3556", "smile...
End of preview. Expand in Data Studio

Chemical OCSR Bench

English | 简体中文

PatentLSBench

Chemical OCSR Bench evaluates molecular recognition from chemical structure images, and chemical-region localization and recognition from single-page patent PDFs. This PatentLSBench dataset provides page and region annotations with complete-structure molecular references.

Data files · Task description · Evaluation protocol

Dataset overview

V1.0 contains 50 complete pages from 50 patent publication numbers.

Resource Count
Single-page raster PDF inputs 50
Chemical regions and annotated crops 536
Original MOL references 277
Valid canonical-SMILES references 272
Valid standard-InChIKey references 272
Label dimensions / dictionary values 9 / 68

Tasks and evaluation

All 50 supplied PDFs contain one page-sized raster image and no embedded text layer. This edition measures performance on these raster-page inputs; it does not establish performance on born-digital PDFs.

Mode Model input Evaluation
Image An annotated chemical-region crop Molecular recognition by full standard InChIKey; precision, recall and F1
PDF A supplied single-page patent PDF Layout by one-to-one IoU ≥ 0.5 matching; molecular recognition by full standard InChIKey; separate precision, recall and F1

Layout covers all 536 annotated chemical regions. Molecular recognition uses the fixed 272-target InChIKey list. The region table retains every chemical region; molecular scoring targets are identified in evaluation_masks.json. Original-MOL, SMILES and InChIKey populations are recorded separately.

Report image recognition, PDF Layout and PDF molecular recognition separately, retaining repeated occurrences. Canonical-SMILES exact match is a supplementary metric on its own fixed list. Joint page recovery requires both the fixed Layout match and correct molecular identity and is reported as recovered targets / fixed molecular targets.

Boxes use normalized [x1, y1, x2, y2] coordinates with a top-left origin. Page images are previews; the PDF task takes the supplied PDF. Molecular parsing preserves components, formal charge, isotopes, explicit hydrogens and specified stereochemistry. See the evaluation protocol and molecular reference protocol for the exact settings.

Feed only PDF bytes in PDF mode and crop image bytes in image mode. Keep layout_gt, ocsr_gt, molecular identifiers and annotation labels out of model requests.

Evaluation: In PDF mode, fix the one-to-one Layout assignment before molecular comparison and reuse it for both SMILES and InChIKey. Non-complete structures are Layout-only; complete structures without usable identifiers are unscored for the corresponding molecular field. Extra, duplicate and invalid nonempty candidates count as false positives; missing eligible targets count as false negatives. See the evaluation protocol for the complete counting rules.

Configurations and files

Each configuration provides a test split.

Configuration Rows Contents
pages · default 50 Page identifiers, PDF paths, nested Layout/OCSR references and embedded page_image previews
regions 536 Region identifiers, bounding boxes, labels, nullable MOL paths and embedded region_image crops
molecules 277 MOL-reference records, nullable SMILES/InChIKey fields and embedded region_image crops

The edition directory public-50-pages contains pages.jsonl, regions.jsonl, ocsr.jsonl, assets/, schema, statistics, fixed target lists and normalization resources. Join regions.page_id = pages.id, molecules.page_id = pages.id and molecules.region_id = regions.id. Asset paths are relative to that edition directory.

The nine label fields are type, chemical_family, component_features, atom_bond_features, stereo_features, notation_features, topology_features, scene and image_defects. Labels support multiple values and retain their Chinese dictionary values. Field definitions, nullable values and counting rules are documented in the schema and edition README.

Quick start

from datasets import load_dataset

pages = load_dataset("PatSnap/chemical-ocsr-bench", "pages", split="test", revision="V1.0")
regions = load_dataset("PatSnap/chemical-ocsr-bench", "regions", split="test", revision="V1.0")
molecules = load_dataset("PatSnap/chemical-ocsr-bench", "molecules", split="test", revision="V1.0")
print(pages[0]["id"], regions[0]["region_image"].size)

Join the views by their actual keys without loading preview images:

page_by_id = {row["id"]: row for row in pages.select_columns(["id", "pdf_path"])}
region_by_id = {row["id"]: row for row in regions.select_columns(["id", "page_id"])}
reference = molecules.select_columns(["page_id", "region_id"])[0]
page = page_by_id[reference["page_id"]]
region = region_by_id[reference["region_id"]]
assert region["page_id"] == page["id"]

Download the PDF inputs and reference assets:

from pathlib import Path
from huggingface_hub import snapshot_download

snapshot = snapshot_download(
    repo_id="PatSnap/chemical-ocsr-bench",
    repo_type="dataset",
    revision="V1.0",
    allow_patterns="public-50-pages/**",
)
edition = Path(snapshot) / "public-50-pages"
pdf = edition / pages[0]["pdf_path"]
pdf_bytes = pdf.read_bytes()

Citation

@dataset{patsnap_chemical_ocsr_bench_2026,
  title   = {Patent Page Chemical Structure Detection and Recognition Benchmark},
  author  = {PatSnap},
  year    = {2026},
  version = {1.0},
  url     = {https://huggingface.co/datasets/PatSnap/chemical-ocsr-bench}
}

License and Disclaimer

The dataset, annotations and accompanying assets use CC BY-NC 4.0. Project documentation uses the MIT License; third-party materials retain their applicable rights and separate terms. See the normalization source notice.

The dataset and related outputs may contain errors or omissions and do not constitute legal, patent, medical or regulatory advice or professional certification. Review and validate them before relying on them; see the full disclaimer.

Downloads last month
63