CLIPViC (CLIP & PViC) for Human-Object Interaction Detection
This is the official implementation code for CLIPViC.
Status
We don't update the code anymore. Please contact quanhutuo@qq.com (emil) with any questions.
| Model | Dataset | Backbone | Default Settings |
|---|---|---|---|
| Ours | HICO-DET | ResNet50+B/32 | (36.70, 34.82, 37.27) |
| Ours | HICO-DET | ResNet50+B/16 | (37.55, 35.38, 38.20) |
| Ours | HICO-DET | Swin-L+L/14@336 | (48.02, 49.97, 47.43) |
| Model | Dataset | Backbone | Scenario 1 | Scenario 2 |
|---|---|---|---|---|
| Ours | V-COCO | ResNet50+B/32 | 60.5 |
66.2 |
| Ours | V-COCO | ResNet50+B/16 | 61.0 |
66.7 |
| Ours | V-COCO | Swin-L+L/14@336 | 63.0 |
69.4 |