Fps is the reported evaluation metric for Total-Text. Codesota tracks published model scores on this metric so readers can compare state-of-the-art results across sources and model families.
Higher is better
| Rank | Model | Trust | Score | Year | Links | Fix |
|---|---|---|---|---|---|---|
| 01 | FAST-T-448 | verified | 152.8 | 2021 | Paper ↗Code ↗ | Looks wrong? |
| 02 | FAST-S-512 | verified | 115.5 | 2021 | Paper ↗Code ↗ | Looks wrong? |
| 03 | FAST-B-512 | verified | 93.2 | 2021 | Paper ↗Code ↗ | Looks wrong? |
Precision is the reported evaluation metric for Total-Text. Codesota tracks published model scores on this metric so readers can compare state-of-the-art results across sources and model families.
Higher is better
F Measure is the reported evaluation metric for Total-Text. Codesota tracks published model scores on this metric so readers can compare state-of-the-art results across sources and model families.
Higher is better
F Measure Full Lexicon is the reported evaluation metric for Total-Text. Codesota tracks published model scores on this metric so readers can compare state-of-the-art results across sources and model families.
Higher is better
| Rank | Model | Trust | Score | Year | Links | Fix |
|---|---|---|---|---|---|---|
| 01 | DeepSolo (ViTAEv2-S, TextOCR) | verified | 89.6 | 2022 | Paper ↗Code ↗ | Looks wrong? |
| 02 | DeepSolo (ResNet-50, TextOCR) | verified | 88.7 | 2022 | Paper ↗Code ↗ | Looks wrong? |
| 03 | DeepSolo (ResNet-50) | verified | 87 | 2022 | Paper ↗Code ↗ | Looks wrong? |
| 04 | UNITS | verified | 86 | 2023 | Paper ↗Code ↗ | Looks wrong? |
| 05 | A3S | verified | 85.1 | 2023 | Paper ↗ | Looks wrong? |
| 06 | SwinTextSpotter | verified | 84.1 | 2022 | Paper ↗Code ↗ | Looks wrong? |
| 07 | TESTR | verified | 83.9 | 2022 | Paper ↗Code ↗ | Looks wrong? |
| 08 | MANGO | verified | 83.6 | 2020 | Paper ↗Code ↗ | Looks wrong? |
| 09 | DEER | verified | 83.3 | 2022 | Paper ↗ | Looks wrong? |
| 10 | GLASS | verified | 83 | 2022 | Paper ↗Code ↗ | Looks wrong? |
| 11 | MaskTextSpotter v3 | verified | 78.4 | 2020 | Paper ↗Code ↗ | Looks wrong? |
| 12 | ABCNet v2 | verified | 78.1 | 2021 | Paper ↗Code ↗ | Looks wrong? |
Recall is the reported evaluation metric for Total-Text. Codesota tracks published model scores on this metric so readers can compare state-of-the-art results across sources and model families.
Higher is better
F Measure No Lexicon is the reported evaluation metric for Total-Text. Codesota tracks published model scores on this metric so readers can compare state-of-the-art results across sources and model families.
Higher is better
| Rank | Model | Trust | Score | Year | Links | Fix |
|---|---|---|---|---|---|---|
| 01 | DeepSolo (ViTAEv2-S, TextOCR) | verified | 83.6 | 2022 | Paper ↗Code ↗ | Looks wrong? |
| 02 | DeepSolo (ResNet-50, TextOCR) | verified | 82.5 | 2022 | Paper ↗Code ↗ | Looks wrong? |
| 03 | DeepSolo (ResNet-50) | verified | 79.7 | 2022 | Paper ↗Code ↗ | Looks wrong? |
| 04 | A3S | verified | 79.4 | 2023 | Paper ↗ | Looks wrong? |
| 05 | UNITS | verified | 78.7 | 2023 | Paper ↗Code ↗ | Looks wrong? |