Text Recognition on Street View Text (SVT)
98.8AccuracyCA-FCN
Evaluation Results
| Method | Links | |
|---|---|---|
| CA-FCNLexicon Size=50, Extra synthetic data=true2018.09 | 98.8 | |
| CA-FCNLexicon Size=50, Extra synthetic data=false2018.09 | 98.5 | |
| CRNNLexicon Size=502018.09 | 97.5 | |
| Cheng et al. [6]Lexicon=502017.11 | 97.1 | |
| FANLexicon Size=502018.09 | 97.1 | |
| Naive_baseLexicon=502017.11 | 96.9 | |
| Bai et al. 2018Lexicon Size=502018.09 | 96.6 | |
| Shi et al. [31]Lexicon=502017.11 | 96.4 | |
| Lee et al. [22]Lexicon=502017.11 | 96.3 | |
| STN_baseLexicon=502017.11 | 96.3 | |
| Lee and Osindero 2016Lexicon Size=502018.09 | 96.3 | |
| Wang and Hu 2017Lexicon Size=502018.09 | 96.3 | |
| Ours (AON)Lexicon=502017.11 | 96 | |
| AONLexicon Size=502018.09 | 96 | |
| Cheng's baseline [6]Lexicon=502017.11 | 95.7 | |
| Shi et al. [32]Lexicon=502017.11 | 95.5 | |
| RARELexicon Size=502018.09 | 95.5 | |
| ProposedLexicon=50-word2014.12 | 95.4 | |
| Jaderberg et al. [17]Lexicon=502017.11 | 95.4 | |
| Jaderberg et al. 2016Lexicon Size=502018.09 | 95.4 | |
| Yang et al. [39]Lexicon=502017.11 | 95.2 | |
| Yang et al. 2017Lexicon Size=502018.09 | 95.2 | |
| Jaderberg et al. [16]Lexicon=50, Row=142017.11 | 93.2 | |
| Jaderberg et al. 2015aLexicon Size=502018.09 | 93.2 | |
| GordoLexicon=502017.11 | 91.8 | |
| Gordo 2015Lexicon Size=502018.09 | 91.8 | |
| GordoLexicon=50-word2014.12 | 90.7 | |
| PhotoOCRLexicon=50-word2014.12 | 90.4 | |
| Bissacco et al.Lexicon=502017.11 | 90.4 | |
| Almazán et al.Lexicon=502017.11 | 89.2 | |
| Almazán et al. 2014Lexicon Size=502018.09 | 89.2 | |
| Bai et al. 2018Lexicon Size=02018.09 | 87.5 | |
| CA-FCNLexicon Size=0, Extra synthetic data=true2018.09 | 86.4 | |
| JaderbergLexicon=50-word2014.12 | 86.1 | |
| Jaderberg et al. [16]Lexicon=None, Row=102017.11 | 86.1 | |
| Jaderberg, Vedaldi, and Zisserman 2014Lexicon Size=502018.09 | 86.1 | |
| Cheng et al. [6]Lexicon=None2017.11 | 85.9 | |
| FANLexicon Size=02018.09 | 85.9 | |
| Su and LuLexicon=502017.11 | 83 | |
| Su and Lu 2014Lexicon Size=502018.09 | 83 | |
| Ours (AON)Lexicon=None2017.11 | 82.8 | |
| AONLexicon Size=02018.09 | 82.8 | |
| CRNNLexicon Size=02018.09 | 82.7 | |
| Cheng's baseline [6]Lexicon=None2017.11 | 82.2 | |
| CA-FCNLexicon Size=0, Extra synthetic data=false2018.09 | 82.1 | |
| Shi et al. [32]Lexicon=None2017.11 | 81.9 | |
| Naive_baseLexicon=None2017.11 | 81.9 | |
| RARELexicon Size=02018.09 | 81.9 | |
| Wang and Hu 2017Lexicon Size=02018.09 | 81.5 | |
| Shi et al. [31]Lexicon=None2017.11 | 80.8 | |
| ProposedLexicon=None2014.12 | 80.7 | |
| Jaderberg et al. [17]Lexicon=None2017.11 | 80.7 | |
| Lee et al. [22]Lexicon=None2017.11 | 80.7 | |
| STN_baseLexicon=None2017.11 | 80.7 | |
| Jaderberg et al. 2016Lexicon Size=02018.09 | 80.7 | |
| Lee and Osindero 2016Lexicon Size=02018.09 | 80.7 | |
| PhotoOCRLexicon=None2014.12 | 78 | |
| Bissacco et al.Lexicon=None2017.11 | 78 | |
| GoelLexicon=50-word2014.12 | 77.3 | |
| Goel et al.Lexicon=502017.11 | 77.3 | |
| YaoLexicon=50-word2014.12 | 75.9 | |
| Yao et al.Lexicon=502017.11 | 75.9 | |
| Yao et al. 2014Lexicon Size=502018.09 | 75.9 | |
| AlsharifLexicon=50-word2014.12 | 74.3 | |
| AlsharifLexicon=502017.11 | 74.3 | |
| MishraLexicon=50-word2014.12 | 73.2 | |
| Mishra et al.Lexicon=502017.11 | 73.2 | |
| Mishra, Alahari, and Jawahar 2012aLexicon Size=502018.09 | 73.2 | |
| NovikovaLexicon=50-word2014.12 | 72.9 | |
| Jaderberg et al. [16]Lexicon=None, Row=142017.11 | 71.7 | |
| Jaderberg et al. 2015aLexicon Size=02018.09 | 71.7 | |
| Wang & WuLexicon=50-word2014.12 | 70 | |
| Wang et al.Lexicon=50, Citation=[37]2017.11 | 70 | |
| Wang et al. 2012Lexicon Size=502018.09 | 70 | |
| Rodríguez-Serrano, Gordo, and Perronnin 2015Lexicon Size=502018.09 | 70 | |
| WangLexicon=50-word2014.12 | 57 | |
| Wang et al.Lexicon=50, Citation=[35]2017.11 | 57 | |
| Wang, Babenko, and Belongie 2011Lexicon Size=502018.09 | 57 | |
| Baseline (ABBYY)Lexicon=50-word2014.12 | 35 | |
| ABBYYLexicon=502017.11 | 35 |