Text Spotting on ICDAR 2013 (test)
93.4E2E Acc (Strong Lexicon)MANGO
Evaluation Results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| MANGOInput Size=L-1440, uses specific lexicons from (Liao et al. 2019)=true2020.12 | 93.4 | 92.3 | 88.7 | 92.9 | 92.7 | 88.3 | 6.3 | |
| Mask TextSpotterInput Size=S-1000, uses specific lexicons from (Liao et al. 2019)=true2020.12 | 93.3 | 91.3 | 88.2 | 92.7 | 91.7 | 87.7 | 3.1 | |
| Text PerceptronInput Size=L-14402020.12 | 91.4 | 90.7 | 85.8 | 94.9 | 94 | 88.5 | — | |
| He et al. †support for irregular text=false2020.12 | 91 | 89 | 86 | 93 | 92 | 87 | — | |
| MANGOInput Size=L-14402020.12 | 90.5 | 90 | 86.9 | 94.8 | 94.1 | 90.1 | 6.3 | |
| TextNetInput Size=L-9202020.12 | 89.8 | 88.9 | 83 | 94.6 | 94.5 | 87 | 2.7 | |
| MANGOInput Size=L-10802020.12 | 89.7 | 89.3 | 85.3 | 94 | 93.4 | 88.4 | 9.8 | |
| FOTS +Input Size=L-920, support for irregular text=false2020.12 | 88.8 | 87.1 | 80.8 | 92.7 | 90.7 | 83.5 | 22 | |
| BoundaryInput Size=L-12802020.12 | 88.2 | 87.7 | 84.1 | — | — | — | — |