Hand Instance Segmentation on HaDR (test)
10.4PDQmaxMask R-CNN
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| Mask R-CNNTraining Dataset=EgoHands, Backbone=ResNet101, Input Modality=RGB2023.04 | 10.4 | 0.9 | 21.4 | 25.3 | |
| SOLOv2Training Dataset=EgoHands, Backbone=ResNet50, Input Modality=RGB2023.04 | 10.25 | 0.575 | 26.4 | 29.9 | |
| SOLOv2Training Dataset=EgoHands, Backbone=ResNet101, Input Modality=RGB2023.04 | 9.75 | 0.5 | 22.9 | 25 | |
| Mask R-CNNTraining Dataset=RHD, Backbone=ResNet101, Input Modality=RGB-D2023.04 | 8.39 | 0.975 | 16.9 | 17.8 | |
| Mask R-CNNTraining Dataset=ObMan, Backbone=ResNet50, Input Modality=RGB-D2023.04 | 7.98 | 0.975 | 18.7 | 21.7 | |
| SOLOv2Training Dataset=RHD, Backbone=ResNet101, Input Modality=RGB-D2023.04 | 7.94 | 0.675 | 15.7 | 16.8 | |
| Mask R-CNNTraining Dataset=ObMan, Backbone=ResNet101, Input Modality=RGB-D2023.04 | 7.86 | 0.95 | 20.6 | 22.7 | |
| Mask R-CNNTraining Dataset=RHD, Backbone=ResNet50, Input Modality=RGB-D2023.04 | 7.75 | 0.975 | 16.5 | 17.5 | |
| Mask R-CNNTraining Dataset=EgoHands, Backbone=ResNet50, Input Modality=RGB2023.04 | 7.42 | 0.95 | 16.1 | 17.4 | |
| SOLOv2Training Dataset=ObMan, Backbone=ResNet50, Input Modality=RGB-D2023.04 | 6.06 | 0.45 | 13.8 | 16.5 | |
| SOLOv2Training Dataset=RHD, Backbone=ResNet50, Input Modality=RGB-D2023.04 | 5.91 | 0.725 | 13 | 14.8 | |
| SOLOv2Training Dataset=ObMan, Backbone=ResNet101, Input Modality=RGB-D2023.04 | 5.35 | 0.375 | 13.5 | 16 | |
| SOLOv2Training Dataset=HandSeg, Backbone=ResNet101, Input Modality=Depth2023.04 | 1.94 | 0.45 | 6 | 6.9 | |
| Mask R-CNNTraining Dataset=HandSeg, Backbone=ResNet101, Input Modality=Depth2023.04 | 1.58 | 0.925 | 1.7 | 1.8 | |
| Mask R-CNNTraining Dataset=HandSeg, Backbone=ResNet50, Input Modality=Depth2023.04 | 1.47 | 0.975 | 1.8 | 2 | |
| SOLOv2Training Dataset=HandSeg, Backbone=ResNet50, Input Modality=Depth2023.04 | 1.04 | 0.35 | 3.5 | 3.9 | |
| Mask R-CNNTraining Dataset=DenseHands, Backbone=ResNet50, Input Modality=Depth2023.04 | 0.61 | 0.975 | 8.4 | 9.3 | |
| Mask R-CNNTraining Dataset=DenseHands, Backbone=ResNet101, Input Modality=Depth2023.04 | 0.34 | 0.975 | 2.7 | 2.9 | |
| SOLOv2Training Dataset=DenseHands, Backbone=ResNet101, Input Modality=Depth2023.04 | 0.01 | 0.225 | 0.8 | 0.8 | |
| SOLOv2Training Dataset=DenseHands, Backbone=ResNet50, Input Modality=Depth2023.04 | 0 | 0 | 0.9 | 0.9 |