Human-Object Interaction Segmentation on HICO-DET (test)
48.6mask mAP (Full)RLIPv2
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| RLIPv2Backbone=Swin-L, Category=Additional training with object detection data2024.08 | 48.6 | 44.3 | 49.8 | |
| FleVRSBackbone=Focal-L, Category=Single-stage methods2024.08 | 40.5 | 34.9 | 42.4 | |
| VIPLOBackbone=ViT-B, Category=Bottom-up methods2024.08 | 39.1 | 37.8 | 39.7 | |
| GEN-VLKTBackbone=ResNet-101, Category=Single-stage methods2024.08 | 35.6 | 32.6 | 37.8 | |
| MURENBackbone=ResNet-50, Category=Single-stage methods2024.08 | 35.4 | 30.1 | 37.6 | |
| UPTBackbone=ResNet-101, Category=Bottom-up methods2024.08 | 34.9 | 29.4 | 36.1 | |
| RLIPBackbone=ResNet-50, Category=Single-stage methods, Training Data=VG+COCO2024.08 | 34.4 | 27.7 | 36.5 | |
| CDNBackbone=ResNet-101, Category=Single-stage methods2024.08 | 33.9 | 28.9 | 36 | |
| SCGBackbone=ResNet-50, Category=Bottom-up methods2024.08 | 31.3 | 25 | 35.5 | |
| STIPBackbone=ResNet-50, Category=Bottom-up methods2024.08 | 30.8 | 28.6 | 32.5 | |
| QPICBackbone=ResNet-101, Category=Single-stage methods2024.08 | 30.5 | 23.1 | 33.1 | |
| HOTRBackbone=ResNet-50, Category=Single-stage methods2024.08 | 26.5 | 18.5 | 29 |