Physical structure recognition on ICDAR cTDaR archival 2019
82.2PrecisionTabstruct-Net
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| Tabstruct-NetTraining Dataset=cTDaR + SCITSR, Number of training images=12.6K, Experimental Setup=S-A2020.10 | 82.2 | 78.7 | 80.4 | |
| DGCNNTraining Dataset=cTDaR + SCITSR, Number of training images=12.6K, Experimental Setup=S-A2020.10 | 80.3 | 77.8 | 79 | |
| Tabstruct-NetTraining Dataset=cTDaR, Number of training images=0.6K, Experimental Setup=S-A2020.10 | 80.3 | 76.8 | 78.5 | |
| SPLERGETraining Dataset=cTDaR + SCITSR, Number of training images=12.6K, Experimental Setup=S-A2020.10 | 79.2 | 80 | 79.6 | |
| DGCNNTraining Dataset=cTDaR, Number of training images=0.6K, Experimental Setup=S-A2020.10 | 78.5 | 75.1 | 76.8 | |
| SPLERGETraining Dataset=cTDaR, Number of training images=0.6K, Experimental Setup=S-A2020.10 | 77.4 | 78.3 | 77.8 | |
| NLPR-PALTraining Dataset=cTDaR, Number of training images=0.6K, Experimental Setup=S-A2020.10 | 72 | 77 | 74.5 | |
| Tabstruct-NetTraining Dataset=SCITSR, Number of training images=12.0K, Experimental Setup=S-A2020.10 | 59.5 | 57.2 | 58.3 | |
| SPLERGETraining Dataset=SCITSR, Number of training images=12.0K, Experimental Setup=S-A2020.10 | 55.9 | 57.2 | 56.5 | |
| DGCNNTraining Dataset=SCITSR, Number of training images=12.0K, Experimental Setup=S-A2020.10 | 55.2 | 51.9 | 53.5 |