Scene Text Recognition on IIIT5K 3,000 samples (test)
99.6Word AccuracyDTrOCR
Evaluation Results
| Method | Links | |
|---|---|---|
| DTrOCRTraining data=Real2023.08 | 99.6 | |
| PARSeqTraining data=Real2023.08 | 99.1 | |
| PARSeq-HParameters=0.6B, Train Data=REBU-Syn2023.12 | 98.93 | |
| PARSeq-SParameters=22M, Train Data=REBU-Syn2023.12 | 98.63 | |
| PARSeq-LTrain Data=REBU-Syn2023.12 | 98.63 | |
| TRBATraining data=Real2023.08 | 98.6 | |
| ABINetTraining data=Real2023.08 | 98.6 | |
| PARSeq-BTrain Data=REBU-Syn2023.12 | 98.53 | |
| DTrOCRTraining data=Synth2023.08 | 98.4 | |
| DiffusionSTRTraining Datasets=MJ and ST2023.06 | 97.3 | |
| DiffusionSTRTraining data=Synth2023.08 | 97.3 | |
| PARSeq2021.09 | 97 | |
| PARSeqTraining Datasets=MJ and ST2023.06 | 97 | |
| PARSeqTraining data=Synth2023.08 | 97 | |
| Diaz2021.09 | 96.8 | |
| MaskOCRBackbone=ViT-L2021.09 | 96.5 | |
| MaskOCR_LARGETraining data=Synth2023.08 | 96.5 | |
| SVTR_LARGETraining data=Synth2023.08 | 96.3 | |
| ABINet2021.09 | 96.2 | |
| ABINetTraining Datasets=MJ and ST2023.06 | 96.2 | |
| ABINetTraining data=Synth2023.08 | 96.2 | |
| SVTR_BASETraining data=Synth2023.08 | 96 | |
| VisionLAN2021.09 | 95.8 | |
| MaskOCRBackbone=ViT-B2021.09 | 95.8 | |
| VisionLANTraining data=Synth2023.08 | 95.8 | |
| MaskOCR_BASETraining data=Synth2023.08 | 95.8 | |
| TextScanner2021.09 | 95.7 | |
| TextScannerTraining data=Synth2023.08 | 95.7 | |
| PREN2D2021.09 | 95.6 | |
| PREN2DTraining data=Synth2023.08 | 95.6 | |
| RobustScanner2021.09 | 95.4 | |
| Bhunia2021.09 | 95.2 | |
| CVAE-Feed.2021.09 | 95.2 | |
| JVSRTraining data=Synth2023.08 | 95.2 | |
| CVAE-FeedTraining data=Synth2023.08 | 95.2 | |
| RCEED2021.09 | 94.9 | |
| SRN2021.09 | 94.8 | |
| SRNTraining data=Synth2023.08 | 94.8 | |
| AutoSTR2021.09 | 94.7 | |
| AutoSTRTraining data=Synth2023.08 | 94.7 | |
| CRNNTraining data=Real2023.08 | 94.6 | |
| PlugNet2021.09 | 94.4 | |
| PlugNetTraining data=Synth2023.08 | 94.4 | |
| STN-CSTR2021.09 | 94.2 | |
| TrOCRBackbone=TrOCR-Large, Training Data=Synthetic + Benchmark2021.09 | 94.1 | |
| TrOCRBackbone=TrOCR-Base, Training Data=Synthetic + Benchmark2021.09 | 93.4 | |
| TRBA2021.09 | 92.1 | |
| TRBATraining Datasets=MJ and ST2023.06 | 92.1 | |
| TRBATraining data=Synth2023.08 | 92.1 | |
| TrOCRBackbone=TrOCR-Large, Training Data=Synthetic2021.09 | 91 | |
| TrOCR_LARGETraining data=Synth2023.08 | 91 | |
| TrOCRBackbone=TrOCR-Base, Training Data=Synthetic2021.09 | 90.1 | |
| TrOCR_BASETraining data=Synth2023.08 | 90.1 | |
| VITSTR-B2021.09 | 88.4 | |
| ViTSTRTraining Datasets=MJ and ST2023.06 | 88.4 | |
| ViTSTR_BASETraining data=Synth2023.08 | 88.4 | |
| CRNN2021.09 | 84.3 | |
| CRNNTraining Datasets=MJ and ST2023.06 | 81.8 | |
| CRNNTraining data=Synth2023.08 | 81.8 |