Scene Text Recognition on IIIT5K (50, 1k, None Lexicons)
99.8Accuracy (50 Lexicon)CA-FCN
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| CA-FCNTraining Data=ST2019.12 | 99.8 | 98.8 | 91.9 | |
| TextScanner-preTraining Data=ST2019.12 | 99.8 | 99 | 92.1 | |
| TextScanner-mutualTraining Data=ST2019.12 | 99.8 | 99.2 | 92.6 | |
| TextScanner+realTraining Data=ST + 90k + real2019.12 | 99.8 | 99.5 | 95.7 | |
| TextScanner+90kTraining Data=ST + 90k2019.12 | 99.7 | 99.1 | 93.9 | |
| AONTraining Data=ST + 90k2019.12 | 99.6 | 98.1 | 87 | |
| ASTERTraining Data=ST + 90k2019.12 | 99.6 | 98.8 | 93.4 | |
| EPTraining Data=ST + 90k2019.12 | 99.5 | 97.9 | 88.3 | |
| SARTraining Data=ST + 90k + real2019.12 | 99.4 | 98.2 | 95 | |
| FANTraining Data=ST + 90k2019.12 | 99.3 | 97.5 | 87.4 | |
| Yang et al.Training Data=90k2019.12 | 97.8 | 96.1 | — | |
| CRNN2015.07 | 97.6 | 94.4 | 78.2 | |
| CRNNTraining Data=90k2019.12 | 97.6 | 94.4 | 78.2 | |
| Jaderberg et al. [22]2015.07 | 97.1 | 92.7 | — | |
| Jaderberg et al. (2016)Training Data=90k2019.12 | 97.1 | 92.7 | — | |
| R2AMTraining Data=90k2019.12 | 96.8 | 94.4 | 78.4 | |
| RARETraining Data=90k2019.12 | 96.2 | 93.8 | 81.9 | |
| Jaderberg et al. [21]2015.07 | 95.5 | 89.6 | — | |
| Jaderberg et al. (2014a)Training Data=90k2019.12 | 95.5 | 89.6 | — | |
| Gordo2015.07 | 93.3 | 86.6 | — | |
| GordoTraining Data=-2019.12 | 93.3 | 86.6 | — | |
| Almazán et al.2015.07 | 91.2 | 82.1 | — | |
| Almazán et al.Training Data=-2019.12 | 91.2 | 82.1 | — | |
| Yao et al.2015.07 | 80.2 | 69.3 | — | |
| StrokeletsTraining Data=-2019.12 | 80.2 | 69.3 | — | |
| Rodriguez-Serrano et al.2015.07 | 76.1 | 57.4 | — | |
| Mishra et al.2015.07 | 64.1 | 57.5 | — | |
| ABBYY2015.07 | 24.3 | — | — |