Scene Text Recognition on IC15
92.7AccuracyCSD-D
Evaluation results
| Method | Links | |
|---|---|---|
| CSD-DData=RBU, Params=110.2 M2025.03 | 92.7 | |
| CSD-BData=RBU, Params=104.9 M2025.03 | 92.6 | |
| CSD-BData=Real, Params=104.9 M2025.03 | 92.5 | |
| CSD-DData=Real, Params=110.2 M2025.03 | 92.4 | |
| CSD-SData=RBU, Params=40.8 M2025.03 | 92.2 | |
| CSD-SData=Real, Params=40.8 M2025.03 | 91.9 | |
| CLIP4STR-LData=RBU, Params=446 M2025.03 | 91.9 | |
| CLIP4STR-HData=RBU, Params=1 B2025.03 | 91.7 | |
| CLIP4STR-BData=Real, Params=158 M2025.03 | 91.4 | |
| CLIP4STR-BData=RBU, Params=158 M2025.03 | 91.4 | |
| CLIP4STR-LData=Real, Params=446 M2025.03 | 91.3 | |
| PARSeqData=Real, Params=22.5 M2025.03 | 90.7 | |
| ABINETData=Real, Params=23.5 M2025.03 | 90.2 | |
| TRBAData=Real, Params=49.6 M2025.03 | 89.8 | |
| VITSTR-SData=Real, Params=21.7 M2025.03 | 88.4 | |
| ABINet-LVstLabeled Datasets=MJ+ST, Unlabeled Datasets=Uber-Text2021.03 | 87.4 | |
| ABINet-LVestLabeled Datasets=MJ+ST, Unlabeled Datasets=Uber-Text2021.03 | 86.9 | |
| SVTR-LBackbone=Large2022.04 | 86.6 | |
| ABINet-LVLabeled Datasets=MJ+ST2021.03 | 86 | |
| ABINetLanguage Mode=Lan-aware2022.04 | 86 | |
| VSTLanguage Mode=Lan-aware2022.04 | 85.4 | |
| SVTR-BBackbone=Base2022.04 | 85.2 | |
| SVTR-SBackbone=Small2022.04 | 84.7 | |
| SRN-LV (Reproduced)Labeled Datasets=MJ+ST2021.03 | 84.2 | |
| SVTR-TBackbone=Tiny2022.04 | 84.1 | |
| ABINet-SVLabeled Datasets=MJ+ST2021.03 | 84 | |
| VisionLANLanguage Mode=Lan-aware2022.04 | 83.7 | |
| PREN2DLanguage Mode=Lan-aware2022.04 | 83 | |
| PARSeqTraining Datasets=MJ and ST2023.06 | 82.9 | |
| SRNLabeled Datasets=MJ+ST2021.03 | 82.7 | |
| SRNLanguage Mode=Lan-aware2022.04 | 82.7 | |
| SRN-SV (Reproduced)Labeled Datasets=MJ+ST2021.03 | 82.6 | |
| VST*Language Mode=Lan-free2022.04 | 82.3 | |
| RCEED2021.07 | 82.2 | |
| DiffusionSTRTraining Datasets=MJ and ST2023.06 | 82.2 | |
| CRNNData=Real, Params=8.5 M2025.03 | 82 | |
| AutoSTRLanguage Mode=Lan-aware2022.04 | 81.8 | |
| ABINet*Language Mode=Lan-free2022.04 | 81.7 | |
| CSTR2021.07 | 81.6 | |
| ASTER2021.07 | 81.6 | |
| Our model2021.07 | 80.2 | |
| SE-ASTER (Ours)lexicon=lexicon-free, training_annotations=word-level2020.05 | 80 | |
| SE-ASTERLabeled Datasets=MJ+ST2021.03 | 80 | |
| SEED2021.07 | 80 | |
| TextscannerLabeled Datasets=MJ+ST2021.03 | 79.4 | |
| PREN*Language Mode=Lan-free2022.04 | 79.2 | |
| SATRN2021.07 | 79 | |
| Yang et al. [53]*lexicon=lexicon-free, training_annotations=word-level and character-level2020.05 | 78.7 | |
| ViTSTRLanguage Mode=Lan-free2022.04 | 78.5 | |
| ASTER baseline reproducedlexicon=lexicon-free, training_annotations=word-level2020.05 | 78.4 | |
| SRN*Language Mode=Lan-free2022.04 | 77.5 | |
| SAMLabeled Datasets=MJ+ST2021.03 | 77.3 | |
| RobustScannerLabeled Datasets=MJ+ST2021.03 | 77.1 | |
| Zhan and Lu 2019Rect=true, 2D=false, character-level annotation=false2019.12 | 76.9 | |
| Zhan et al. [57]lexicon=lexicon-free, training_annotations=word-level2020.05 | 76.9 | |
| ParallelLabeled Datasets=MJ+ST2021.03 | 76.3 | |
| Shi et al. 2018Rect=true, 2D=false, character-level annotation=false2019.12 | 76.1 | |
| ASTER [45]lexicon=lexicon-free, training_annotations=word-level2020.05 | 76.1 | |
| ASTERLanguage Mode=Lan-aware2022.04 | 76.1 | |
| Luo et al. [32]lexicon=lexicon-free, training_annotations=word-level2020.05 | 74.7 | |
| TRBATraining Datasets=MJ and ST2023.06 | 74.7 | |
| DAN-2DRect=false, 2D=true, character-level annotation=false2019.12 | 74.5 | |
| DANLabeled Datasets=MJ+ST2021.03 | 74.5 | |
| Liu et al. [28]*lexicon=lexicon-free, training_annotations=word-level and character-level2020.05 | 74.2 | |
| Bai et al. 2018Rect=true, 2D=false, character-level annotation=true2019.12 | 73.9 | |
| Bai et al. [2]*lexicon=lexicon-free, training_annotations=word-level and character-level2020.05 | 73.9 | |
| ViTSTRTraining Datasets=MJ and ST2023.06 | 72.6 | |
| DAN-1DRect=false, 2D=false, character-level annotation=false2019.12 | 71.8 | |
| Baek et al.2021.07 | 71.8 | |
| Fang et al. 2018Rect=true, 2D=false, character-level annotation=false2019.12 | 71.2 | |
| RosettaLanguage Mode=Lan-free2022.04 | 71.2 | |
| Cheng et al. 2017Rect=true, 2D=false, character-level annotation=true2019.12 | 70.6 | |
| Cheng et al. [7]*lexicon=lexicon-free, training_annotations=word-level and character-level2020.05 | 70.6 | |
| CRNNLanguage Mode=Lan-free2022.04 | 69.4 | |
| Li et al. 2019Rect=false, 2D=true, character-level annotation=false2019.12 | 69.2 | |
| Li et al. [23]lexicon=lexicon-free, training_annotations=word-level2020.05 | 69.2 | |
| SARLanguage Mode=Lan-aware2022.04 | 69.2 | |
| Xie et al. 2019Rect=false, 2D=true, character-level annotation=false2019.12 | 68.9 | |
| Xie et al. [52]lexicon=lexicon-free, training_annotations=word-level2020.05 | 68.9 | |
| MORANLexicon=None2019.01 | 68.8 | |
| Luo, Jin, and Sun 2019Rect=true, 2D=true, character-level annotation=false2019.12 | 68.8 | |
| Cheng et al. [6]Lexicon=None2019.01 | 68.2 | |
| Cheng et al. 2018Rect=true, 2D=false, character-level annotation=false2019.12 | 68.2 | |
| Cheng et al. [8]lexicon=lexicon-free, training_annotations=word-level2020.05 | 68.2 | |
| Cheng et al. [5]Lexicon=None2019.01 | 66.2 | |
| CRNNTraining Datasets=MJ and ST2023.06 | 60.4 | |
| DOTA + CRFRefinement=CRF2026.03 | 58.26 | |
| DOTA2026.03 | 58.02 | |
| RES50-DEF(L4)-ViT-AdaptiveDeformable Layer=L4, Config=Adaptive2026.03 | 57.2 | |
| RES50-ViT-PEConfig=Positional Encoding2026.03 | 57.05 | |
| RES50-ATT-ViT2026.03 | 53.88 | |
| RES50-ViTIteration=22026.03 | 53.01 | |
| RES50-ATT-AdaptiveConfig=Adaptive2026.03 | 51.32 | |
| RES50-ViT2026.03 | 51.08 | |
| RES50-ATTIteration=22026.03 | 50.51 | |
| RES50-DEF-(L3–L4)Deformable Layers=L3-L42026.03 | 50.22 | |
| ResNext2026.03 | 47.47 | |
| RES50-ATT2026.03 | 42.9 |