Human-Object Interaction Detection on HICO-DET Unseen Composition (UC)
38.17Unseen APEZ-HOI
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| EZ-HOISetting=UO, Backbone=Resnet50+ViT-L, Trainable Parameters=14.07M, Object Detector Pre-training=HICO-DET2024.10 | 38.17 | 36.02 | 36.38 | |
| EZ-HOISetting=UO, Backbone=Resnet50+ViT-B, Trainable Parameters=6.85M, Object Detector Pre-training=HICO-DET2024.10 | 33.28 | 32.06 | 32.27 | |
| EZ-HOISetting=UO, Backbone=Resnet50+ViT-L, Trainable Parameters=14.07M, Object Detector Pre-training=MS-COCO2024.10 | 33.08 | 31.09 | 31.42 | |
| CLIP4HOISetting=UO, Backbone=Resnet50+ViT-B, Trainable Parameters=56.7M (estimated), Object Detector Pre-training=MS-COCO2024.10 | 31.79 | 32.73 | 32.58 | |
| EZ-HOISetting=UO, Backbone=Resnet50+ViT-B, Trainable Parameters=6.85M, Object Detector Pre-training=MS-COCO2024.10 | 31.63 | 27.16 | 27.9 | |
| UniHOISetting=UO, Backbone=Resnet50+ViT-L, Trainable Parameters=52.3M (estimated), Object Detector Pre-training=MS-COCO2024.10 | 19.72 | 34.76 | 31.56 | |
| ATLType=UC, Zero-shot=true2023.11 | 16.99 | 20.51 | 19.81 | |
| HOICLIPSetting=UO, Backbone=Resnet50+ViT-B, Trainable Parameters=66.18M, Object Detector Pre-training=MS-COCO2024.10 | 16.2 | 30.99 | 28.53 | |
| FCLSetting=UO, Backbone=Resnet50, Object Detector Pre-training=MS-COCO2024.10 | 15.54 | 20.74 | 19.87 | |
| ATLSetting=UO, Backbone=Resnet50, Object Detector Pre-training=MS-COCO2024.10 | 15.11 | 21.54 | 20.47 | |
| FleVRSevaluation_mode=zero-shot2024.08 | 14.48 | 35.28 | 31.71 | |
| FGType=UC, Zero-shot=true2023.11 | 10.93 | 12.6 | 12.26 | |
| GEN-VLKTevaluation_mode=zero-shot2024.08 | 10.51 | 28.92 | 25.63 | |
| GEN-VLKTSetting=UO, Backbone=Resnet50, Trainable Parameters=42.05M, Object Detector Pre-training=MS-COCO2024.10 | 10.51 | 28.92 | 25.63 | |
| Shen et al.Type=UC, Zero-shot=true2023.11 | 5.62 | — | 6.26 | |
| ATLevaluation_mode=zero-shot2024.08 | 5.05 | 14.69 | 13.08 | |
| FCLevaluation_mode=zero-shot2024.08 | 0 | 13.71 | 11.43 |