Scene Text Recognition on IIIT5k
99.8AccuracyCA-FCN
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| CA-FCNLexicon Size=50, Extra synthetic data=false2018.09 | 99.8 | — | |
| CA-FCNLexicon Size=50, Extra synthetic data=true2018.09 | 99.8 | — | |
| Ours (AON)Lexicon=502017.11 | 99.6 | — | |
| AONLexicon Size=502018.09 | 99.6 | — | |
| Naive_baseLexicon=502017.11 | 99.5 | — | |
| STN_baseLexicon=502017.11 | 99.5 | — | |
| Bai et al. 2018Lexicon Size=502018.09 | 99.5 | — | |
| Cheng et al. [6]Lexicon=502017.11 | 99.3 | — | |
| FANLexicon Size=502018.09 | 99.3 | — | |
| Cheng's baseline [6]Lexicon=502017.11 | 98.9 | — | |
| CA-FCNLexicon Size=1k, Extra synthetic data=false2018.09 | 98.9 | — | |
| CA-FCNLexicon Size=1k, Extra synthetic data=true2018.09 | 98.8 | — | |
| Naive_baseLexicon=1k2017.11 | 98.1 | — | |
| Ours (AON)Lexicon=1k2017.11 | 98.1 | — | |
| AONLexicon Size=1k2018.09 | 98.1 | — | |
| Wang and Hu 2017Lexicon Size=502018.09 | 98 | — | |
| Bai et al. 2018Lexicon Size=1k2018.09 | 97.9 | — | |
| Yang et al. [39]Lexicon=502017.11 | 97.8 | — | |
| STN_baseLexicon=1k2017.11 | 97.8 | — | |
| CRNNLexicon Size=502018.09 | 97.8 | — | |
| Yang et al. 2017Lexicon Size=502018.09 | 97.8 | — | |
| Shi et al. [31]Lexicon=502017.11 | 97.6 | — | |
| Cheng et al. [6]Lexicon=1k2017.11 | 97.5 | — | |
| FANLexicon Size=1k2018.09 | 97.5 | — | |
| Jaderberg et al. [17]Lexicon=502017.11 | 97.1 | — | |
| Jaderberg et al. 2016Lexicon Size=502018.09 | 97.1 | — | |
| Lee et al. [22]Lexicon=502017.11 | 96.8 | — | |
| Cheng's baseline [6]Lexicon=1k2017.11 | 96.8 | — | |
| Lee and Osindero 2016Lexicon Size=502018.09 | 96.8 | — | |
| VSTLanguage Mode=Lan-aware2022.04 | 96.3 | — | |
| SVTR-LBackbone=Large2022.04 | 96.3 | — | |
| Shi et al. [32]Lexicon=502017.11 | 96.2 | — | |
| RARELexicon Size=502018.09 | 96.2 | — | |
| ABINetLanguage Mode=Lan-aware2022.04 | 96.2 | — | |
| Yang et al. [39]Lexicon=1k2017.11 | 96.1 | — | |
| Yang et al. 2017Lexicon Size=1k2018.09 | 96.1 | — | |
| SVTR-BBackbone=Base2022.04 | 96 | — | |
| VisionLANLanguage Mode=Lan-aware2022.04 | 95.8 | — | |
| Wang and Hu 2017Lexicon Size=1k2018.09 | 95.6 | — | |
| VST*Language Mode=Lan-free2022.04 | 95.6 | — | |
| PREN2DLanguage Mode=Lan-aware2022.04 | 95.6 | — | |
| Jaderberg et al. [16]Lexicon=50, Row=142017.11 | 95.5 | — | |
| Jaderberg et al. 2015aLexicon Size=502018.09 | 95.5 | — | |
| CRNNLexicon Size=1k2018.09 | 95 | — | |
| SVTR-SBackbone=Small2022.04 | 95 | — | |
| RCEED2021.07 | 94.9 | — | |
| SRNLanguage Mode=Lan-aware2022.04 | 94.8 | — | |
| AutoSTRLanguage Mode=Lan-aware2022.04 | 94.7 | — | |
| ABINet*Language Mode=Lan-free2022.04 | 94.6 | — | |
| Shi et al. [31]Lexicon=1k2017.11 | 94.4 | — | |
| Lee et al. [22]Lexicon=1k2017.11 | 94.4 | — | |
| Lee and Osindero 2016Lexicon Size=1k2018.09 | 94.4 | — | |
| Yang et al. [53]*lexicon=lexicon-free, training_annotations=word-level and character-level2020.05 | 94.4 | — | |
| SVTR-TBackbone=Tiny2022.04 | 94.4 | — | |
| DAN-2DRect=false, 2D=true, character-level annotation=false2019.12 | 94.3 | — | |
| DANVenue=AAAI’20, Training Set=MJ+ST, RNN=Y, FPS=-2022.04 | 94.3 | — | |
| TextScanner*Venue=AAAI’20, Training Set=MJ+ST+Extra, RNN=N, FPS=-2022.04 | 93.9 | — | |
| Shi et al. [32]Lexicon=1k2017.11 | 93.8 | — | |
| RARELexicon Size=1k2018.09 | 93.8 | — | |
| SE-ASTER (Ours)lexicon=lexicon-free, training_annotations=word-level2020.05 | 93.8 | — | |
| SEED2021.07 | 93.8 | — | |
| SEEDVenue=CVPR’20, Training Set=MJ+ST, RNN=Y, FPS=-2022.04 | 93.8 | — | |
| SCATTERVenue=CVPR’20, Training Set=MJ+ST+Extra, RNN=Y, FPS=-2022.04 | 93.7 | — | |
| ASTER baseline reproducedlexicon=lexicon-free, training_annotations=word-level2020.05 | 93.5 | — | |
| Our model2021.07 | 93.5 | — | |
| Shi et al. 2018Rect=true, 2D=false, character-level annotation=false2019.12 | 93.4 | — | |
| ASTER [45]lexicon=lexicon-free, training_annotations=word-level2020.05 | 93.4 | — | |
| ASTERLanguage Mode=Lan-aware2022.04 | 93.4 | — | |
| GordoLexicon=502017.11 | 93.3 | — | |
| Gordo 2015Lexicon Size=502018.09 | 93.3 | — | |
| Zhan and Lu 2019Rect=true, 2D=false, character-level annotation=false2019.12 | 93.3 | — | |
| DAN-1DRect=false, 2D=false, character-level annotation=false2019.12 | 93.3 | — | |
| Zhan et al. [57]lexicon=lexicon-free, training_annotations=word-level2020.05 | 93.3 | — | |
| ESIRVenue=CVPR’19, Training Set=MJ+ST, RNN=Y, FPS=-2022.04 | 93.3 | — | |
| SATRN2021.07 | 92.8 | — | |
| Jaderberg et al. [17]Lexicon=1k2017.11 | 92.7 | — | |
| Jaderberg et al. 2016Lexicon Size=1k2018.09 | 92.7 | — | |
| SRN*Language Mode=Lan-free2022.04 | 92.3 | — | |
| PREN*Language Mode=Lan-free2022.04 | 92.1 | — | |
| CA-FCNLexicon Size=0, Extra synthetic data=false2018.09 | 92 | — | |
| Liao et al. 2019Rect=false, 2D=true, character-level annotation=true2019.12 | 92 | — | |
| Liu et al. [28]*lexicon=lexicon-free, training_annotations=word-level and character-level2020.05 | 92 | — | |
| CA-FCN*Venue=AAAI’19, Training Set=ST, RNN=N, FPS=452022.04 | 92 | — | |
| CA-FCNLexicon Size=0, Extra synthetic data=true2018.09 | 91.9 | — | |
| Liao et al. [25]*lexicon=lexicon-free, training_annotations=word-level and character-level2020.05 | 91.9 | — | |
| Ours-LargeVenue=-, Training Set=MJ+ST, RNN=N, FPS=66.91/255+2022.04 | 91.9 | — | |
| Li et al. 2019Rect=false, 2D=true, character-level annotation=false2019.12 | 91.5 | — | |
| Li et al. [23]lexicon=lexicon-free, training_annotations=word-level2020.05 | 91.5 | — | |
| SARLanguage Mode=Lan-aware2022.04 | 91.5 | — | |
| SARVenue=AAAI’19, Training Set=MJ+ST, RNN=Y, FPS=-2022.04 | 91.5 | — | |
| Almazán et al.Lexicon=502017.11 | 91.2 | — | |
| Almazán et al. 2014Lexicon Size=502018.09 | 91.2 | — | |
| Luo, Jin, and Sun 2019Rect=true, 2D=true, character-level annotation=false2019.12 | 91.2 | — | |
| Luo et al. [32]lexicon=lexicon-free, training_annotations=word-level2020.05 | 91.2 | — | |
| MORANLanguage Mode=Lan-aware2022.04 | 91.2 | — | |
| CSTR2021.07 | 90.1 | — | |
| ASTER2021.07 | 90.1 | — | |
| Jaderberg et al. [16]Lexicon=1k, Row=142017.11 | 89.6 | — | |
| Jaderberg et al. 2015aLexicon Size=1k2018.09 | 89.6 | — | |
| Liu et al. 2018Rect=true, 2D=false, character-level annotation=false2019.12 | 89.4 | — |