Table Detection on ICDAR 2013 (test)
100RecallCDEC-Net
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| CDEC-NetTraining Dataset=D1, Training #Image=4808, IoU=0.52020.08 | 100 | 100 | 100 | 100 | |
| GODTraining Dataset=Marmot, Training #Image=2K, IoU=0.52020.08 | 100 | 98.2 | 99.1 | — | |
| CDEC-NetTraining Dataset=Marmot, Training #Image=2K, IoU=0.52020.08 | 100 | 98.1 | 99.1 | 99.5 | |
| CDEC-NetTraining Dataset=ICDAR-2017, Training #Image=1.6K, IoU=0.52020.08 | 100 | 100 | 100 | 100 | |
| DECNTTraining Dataset=D1, Training #Image=4808, IoU=0.52020.08 | 99.6 | 99.6 | 99.6 | — | |
| Kavasidis et al.Training Dataset=Custom dataset, Training #Image=45K, IoU=0.52020.08 | 98.1 | 97.5 | 97.8 | — | |
| CDEC-NetTraining Dataset=PubLayNet, Training #Image=340K, Fine-tuning Dataset=ICDAR-2013, Fine-tuning #Image=170, IoU=0.52020.08 | 96.8 | 98.7 | 97.7 | 95.9 | |
| F-RCNNTraining Dataset=PubLayNet, Training #Image=340K, Fine-tuning Dataset=ICDAR-2013, Fine-tuning #Image=170, IoU=0.52020.08 | 96.4 | 97.2 | 96.8 | — | |
| Tran et al.IoU=0.52020.08 | 96.4 | 95.2 | 95.8 | — | |
| Tran et al2020.01 | 96.36 | 95.21 | 95.78 | — | |
| TableNet + Semantic Featuressemantic_features=true, fine-tuned=ICDAR 2013, threshold=0.992020.01 | 96.28 | 96.97 | 96.62 | — | |
| TableNet + Semantic Featuressemantic_features=true, threshold=0.992020.01 | 96.21 | 95.47 | 95.83 | — | |
| DeepDeSRT2020.01 | 96.15 | 97.4 | 96.77 | — | |
| M-RCNNTraining Dataset=PubLayNet, Training #Image=340K, Fine-tuning Dataset=ICDAR-2013, Fine-tuning #Image=170, IoU=0.52020.08 | 95.5 | 94 | 94.7 | — | |
| TableNetthreshold=0.992020.01 | 95.01 | 95.47 | 95.47 | — | |
| YOLOv3+A+PGTraining Dataset=ICDAR-2017, Training #Image=1.6K, IoU=0.52020.08 | 94.9 | 100 | 97.3 | — | |
| CDEC-Net+Training Dataset=IIIT-AR-13K, Training #Image=9K, IoU=0.52020.08 | 94.2 | 99.3 | 96.8 | 94.2 | |
| PFTDIoU=0.52020.08 | 91.5 | 93.9 | 92.6 | — |