Scene Text Recognition on SVT 647 (test)
98.76AccuracyPARSeq-B
Evaluation results
| Method | Links | ||
|---|---|---|---|
| PARSeq-BTrain Data=REBU-Syn2023.12 | 98.76 | — | |
| PARSeq-LTrain Data=REBU-Syn2023.12 | 98.61 | — | |
| PARSeq-SParameters=22M, Train Data=REBU-Syn2023.12 | 98.45 | — | |
| PARSeq-HParameters=0.6B, Train Data=REBU-Syn2023.12 | 98.45 | — | |
| PARSeq_ATrain data=R2022.07 | 97.9 | — | |
| ABINetTrain data=R2022.07 | 97.8 | — | |
| PARSeq_NTrain data=R2022.07 | 97.5 | — | |
| ABINet++ (LV)Labeled Datasets=MJ+ST+Real, Vision Model Scale=LV2022.11 | 97.1 | — | |
| TRBATrain data=R2022.07 | 97 | — | |
| TrOCRBackbone=TrOCR-Large, Training Data=Synthetic + Benchmark2021.09 | 96.1 | — | |
| VITSTR-STrain data=R2022.07 | 95.8 | — | |
| ABINet++ (LV)Labeled Datasets=MJ+ST, Vision Model Scale=LV2022.11 | 95.7 | 33.9 | |
| ABINet++‡ (LV)Labeled Datasets=MJ+ST, Unlabeled Datasets=Uber-Text, Vision Model Scale=LV, Training Strategy=ensemble self-training2022.11 | 95.5 | — | |
| PIMNet Qiao et al.Year=2021, Labeled Datasets=MJ+ST+Real2022.11 | 95.4 | — | |
| TrOCRBackbone=TrOCR-Base, Training Data=Synthetic + Benchmark2021.09 | 95.2 | — | |
| ABINet++† (LV)Labeled Datasets=MJ+ST, Unlabeled Datasets=Uber-Text, Vision Model Scale=LV, Training Strategy=self-training2022.11 | 94.9 | — | |
| MaskOCRBackbone=ViT-B2021.09 | 94.7 | — | |
| Diaz2021.09 | 94.6 | — | |
| MaskOCRBackbone=ViT-L2021.09 | 94.1 | — | |
| PREN2DTrain data=S2022.07 | 94 | — | |
| PREN2D2021.09 | 94 | — | |
| PARSeq_ATrain data=S2022.07 | 93.6 | — | |
| PARSeq2021.09 | 93.6 | — | |
| ABINetTrain data=S32022.07 | 93.5 | — | |
| ABINet2021.09 | 93.5 | — | |
| ABINetTrain data=S2022.07 | 93.4 | — | |
| TrOCRBackbone=TrOCR-Large, Training Data=Synthetic2021.09 | 93.2 | — | |
| ABINet++ (SV)Labeled Datasets=MJ+ST, Vision Model Scale=SV2022.11 | 93.2 | 31.6 | |
| GTC Hu et al.Year=2020, Labeled Datasets=MJ+ST+Real2022.11 | 92.9 | — | |
| TRBATrain data=S2022.07 | 92.8 | — | |
| TextScannerTrain data=S*2022.07 | 92.7 | — | |
| TextScanner2021.09 | 92.7 | — | |
| Textscanner Wan et al.Year=2020, Labeled Datasets=MJ+ST+Real2022.11 | 92.7 | — | |
| PARSeq_NTrain data=S2022.07 | 92.6 | — | |
| STN-CSTRTraining Data=MJ+ST, Annotations=word, STN=true2021.02 | 92.3 | — | |
| PlugNetTrain data=S2022.07 | 92.3 | — | |
| STN-CSTRTrain data=S2022.07 | 92.3 | — | |
| PlugNet2021.09 | 92.3 | — | |
| STN-CSTR2021.09 | 92.3 | — | |
| SRN* (LV)Labeled Datasets=MJ+ST, Reproduced=true, Vision Model Scale=LV2022.11 | 92.3 | 26.9 | |
| Bhunia et al.Train data=S2022.07 | 92.2 | — | |
| Bhunia2021.09 | 92.2 | — | |
| Bhunia et al.Year=2021, Labeled Datasets=MJ+ST2022.11 | 92.2 | — | |
| RCEEDTrain data=S,B2022.07 | 91.8 | — | |
| RCEED2021.09 | 91.8 | — | |
| VisionLANTrain data=S2022.07 | 91.7 | — | |
| VITSTR-STrain data=S2022.07 | 91.7 | — | |
| VisionLAN2021.09 | 91.7 | — | |
| VisionLan Wang et al.Year=2021, Labeled Datasets=MJ+ST2022.11 | 91.7 | 11.5 | |
| SRNTraining Data=MJ+ST, Annotations=word2021.02 | 91.5 | — | |
| SRNTrain data=S2022.07 | 91.5 | — | |
| SRN2021.09 | 91.5 | — | |
| SRN Yu et al.Year=2020, Labeled Datasets=MJ+ST2022.11 | 91.5 | 46.2 | |
| SATRNTraining Data=MJ+ST, Annotations=word2021.02 | 91.3 | — | |
| PIMNet Qiao et al.Year=2021, Labeled Datasets=MJ+ST2022.11 | 91.2 | 19.1 | |
| TrOCRBackbone=TrOCR-Base, Training Data=Synthetic2021.09 | 91 | — | |
| AutoSTRTrain data=S2022.07 | 90.9 | — | |
| AutoSTR2021.09 | 90.9 | — | |
| SRN* (SV)Labeled Datasets=MJ+ST, Reproduced=true, Vision Model Scale=SV2022.11 | 90.9 | 24.2 | |
| CRNNTrain data=R2022.07 | 90.7 | — | |
| TextScannerTraining Data=MJ+ST, Annotations=word,char2021.02 | 90.1 | — | |
| CSTRTraining Data=MJ+ST, Annotations=word, STN=false2021.02 | 90.1 | — | |
| Textscanner Wan et al.Year=2020, Labeled Datasets=MJ+ST2022.11 | 90.1 | — | |
| SEEDTraining Data=MJ+ST, Annotations=word, Beam search=true2021.02 | 89.6 | — | |
| SEED Qiao et al.Year=2020, Labeled Datasets=MJ+ST2022.11 | 89.6 | 53.3 | |
| ASTERTraining Data=MJ+ST, Annotations=word, Beam search=true2021.02 | 89.5 | — | |
| Robust ScannerTrain data=S,B2022.07 | 89.3 | — | |
| RobustScanner2021.09 | 89.3 | — | |
| DANTraining Data=MJ+ST, Annotations=word2021.02 | 89.2 | — | |
| DAN Wang et al.Year=2020, Labeled Datasets=MJ+ST2022.11 | 89.2 | — | |
| TRBATrain data=S2022.07 | 88.9 | — | |
| TRBA2021.09 | 88.9 | — | |
| RobustScanner Yue et al.Year=2020, Labeled Datasets=MJ+ST2022.11 | 88.1 | 36 | |
| VITSTR-BTrain data=S22022.07 | 87.7 | — | |
| VITSTR-B2021.09 | 87.7 | — | |
| EPSetting=Reported results, Year=2018, Train data=MJ+ST2019.04 | 87.5 | — | |
| Our best modelSetting=Our experiment, Train data=MJ+ST2019.04 | 87.5 | — | |
| Baek et al.Training Data=MJ+ST, Annotations=word2021.02 | 87.5 | — | |
| SSFLSetting=Reported results, Year=2018, Train data=MJ2019.04 | 87.1 | — | |
| STAR-NetSetting=Our experiment, Year=2016, Train data=MJ+ST2019.04 | 86.9 | — | |
| CA-FCNTraining Data=ST, Annotations=word,char2021.02 | 86.4 | — | |
| FANSetting=Reported results, Year=2017, Train data=MJ+ST+C2019.04 | 85.9 | — | |
| FANTraining Data=MJ+ST, Annotations=word2021.02 | 85.9 | — | |
| RARESetting=Our experiment, Year=2016, Train data=MJ+ST2019.04 | 85.8 | — | |
| CRNNTrain data=S2022.07 | 85.7 | — | |
| RosettaSetting=Our experiment, Year=2018, Train data=MJ+ST2019.04 | 84.7 | — | |
| SARTraining Data=MJ+ST, Annotations=word, Test-Time Augmentation (TTA)=true2021.02 | 84.5 | — | |
| Char-NetSetting=Reported results, Year=2018, Train data=MJ2019.04 | 84.4 | — | |
| GRCNNSetting=Our experiment, Year=2017, Train data=MJ+ST2019.04 | 83.7 | — | |
| STAR-NetSetting=Reported results, Year=2016, Train data=MJ+PRI2019.04 | 83.6 | — | |
| AONSetting=Reported results, Year=2018, Train data=MJ+ST2019.04 | 82.8 | — | |
| AONTraining Data=MJ+ST, Annotations=word2021.02 | 82.8 | — | |
| R2AMSetting=Our experiment, Year=2016, Train data=MJ+ST2019.04 | 82.4 | — | |
| RARESetting=Reported results, Year=2016, Train data=MJ2019.04 | 81.9 | — | |
| CRNNSetting=Our experiment, Year=2015, Train data=MJ+ST2019.04 | 81.6 | — | |
| GRCNNSetting=Reported results, Year=2017, Train data=MJ2019.04 | 81.5 | — | |
| CRNNSetting=Reported results, Year=2015, Train data=MJ2019.04 | 80.8 | — | |
| R2AMSetting=Reported results, Year=2016, Train data=MJ2019.04 | 80.7 | — | |
| CRNNTrain data=S2022.07 | 78.9 | — | |
| CRNN2021.09 | 78.9 | — |