Table Detection on ICDAR cTDaR 2019 (archival)
0.981IoU@0.6BEIT-B
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| BEIT-BBackbone=BEIT-B2022.03 | 0.981 | 0.981 | 0.9582 | 0.943 | 0.9635 | |
| DeiT-BBackbone=DeiT-B2022.03 | 0.9754 | 0.9716 | 0.9641 | 0.9263 | 0.9568 | |
| MAE-BBackbone=MAE-B2022.03 | 0.9754 | 0.9754 | 0.9603 | 0.9414 | 0.9612 | |
| DiT-BBackbone=DiT-B2022.03 | 0.9753 | 0.9715 | 0.9602 | 0.9488 | 0.9624 | |
| DiT-LBackbone=DiT-L2022.03 | 0.9753 | 0.9715 | 0.9639 | 0.9526 | 0.9646 | |
| DiT-L (Cascade)Backbone=DiT-L, Detection Framework=Cascade2022.03 | 0.9734 | 0.9734 | 0.9734 | 0.962 | 0.97 | |
| 1st place in cTDaR2022.03 | 0.9716 | 0.9641 | 0.9527 | 0.9112 | 0.9467 | |
| DiT-B (Cascade)Backbone=DiT-B, Detection Framework=Cascade2022.03 | 0.9697 | 0.9697 | 0.9697 | 0.9583 | 0.9663 | |
| ResNeXt-101-32x8d (Cascade)Backbone=ResNeXt-101-32x8d, Detection Framework=Cascade2022.03 | 0.9676 | 0.9638 | 0.9524 | 0.9371 | 0.9535 | |
| ResNeXt-101-32x8dBackbone=ResNeXt-101-32x8d2022.03 | 0.966 | 0.966 | 0.9509 | 0.917 | 0.9473 |