Research map / Detection and segmentation / Open-set and few-shot detection
Visual Compositional Learning for Human-Object Interaction Detection
Zhi Hou, Xiaojiang Peng, Yu Qiao, Dacheng Tao
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add Visual Compositional Learning (VCL) (ablation) | HOI detector · HICO-DET Full | mAP: 18.43 -> 19.43 | Improved |
| add VCL (ablation) | HOI detector · HICO-DET Rare | mAP: 14.14 -> 16.55 | Improved |
| add VCL (ablation) | HOI detector · V-COCO | AP_role: 47.5 -> 48.3 | Improved |
| add VCL in zero-shot rare-first (ablation) | HOI detector · HICO-DET unseen (rare first) | mAP: 3.30 -> 7.55 | Improved |
| add VCL in zero-shot non-rare-first (ablation) | HOI detector · HICO-DET unseen (non-rare first) | mAP: 5.06 -> 9.13 | Improved |
| learn verb representation from union box instead of human box (ablation) | VCL · HICO-DET Full | mAP: 18.57 -> 19.43 | Improved |
| learn verb representation from union box (ablation) | VCL · V-COCO | AP_role: 47.7 -> 48.3 | Improved |
| share ResNet block weights between human and verb stream (ablation) | VCL · HICO-DET Full | mAP: 18.93 -> 19.43 | Improved |
| add re-weighting (ablation) | VCL · HICO-DET Full | mAP: 17.35 -> 19.43 | Improved |
| add Spatial-Human branch at inference (ablation) | VCL · HICO-DET Full | mAP: 16.89 -> 19.43 | Improved |
| add Spatial-Human branch to training (verb-object branch evaluated alone) (ablation) | VCL · HICO-DET Full | mAP: 15.77 -> 16.89 | Improved |
| compose interactions between images (vs within images) (ablation) | VCL · HICO-DET Full | mAP: 18.43 -> 19.06 | Improved |
Closest papers by meaning
- Detecting Human-Object Interaction via Fabricated Compositional Learning (CVPR 2021 · similarity 0.95)
- LINK: Learning Instance-level Knowledge from Vision-Language Models for Human-Object Interaction Detection (ICLR 2026 · similarity 0.91)
- Discovering Syntactic Interaction Clues for Human-Object Interaction Detection (CVPR 2024 · similarity 0.90)
- Toward Open-Set Human Object Interaction Detection (AAAI 2024 · similarity 0.89)
- Human-Object Interaction Detection Collaborated with Large Relation-driven Diffusion Models (NeurIPS 2024 · similarity 0.89)
- Exploring Pose-Aware Human-Object Interaction via Hybrid Learning (CVPR 2024 · similarity 0.89)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
