Research map / Detection and segmentation / Open-set and few-shot detection
Visual Relation Diffusion for Human-Object Interaction Detection
Ping Cao, Yepeng Tang, Chunjie Zhang, Xiaolong Zheng, Chao Liang, Yunchao Wei, Yao Zhao
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 Relation Encoding (VRE) module (ablation) | VRDiff baseline · HICO-DET UV | unseen mAP: 22.26 -> 24.81 | Improved |
| add Spatial Interaction Embedding (SIE) on top of VRE (ablation) | VRDiff · HICO-DET UV | unseen mAP: 24.81 -> 25.99 | Improved |
| add Stable Diffusion feedback on top of VRE+SIE (ablation) | VRDiff · HICO-DET UV | unseen mAP: 25.99 -> 32.14 | Improved |
| add Stable Diffusion feedback on top of VRE+SIE (ablation) | VRDiff · HICO-DET UV | full mAP: 38.15 -> 39.43 | Improved |
| use empty text plus visual condition instead of real interaction text (ablation) | VRDiff · HICO-DET UV | full mAP: 38.19 -> 39.43 | Improved |
| drop real text from RealText+Visual condition (ablation) | VRDiff · HICO-DET UV | full mAP: 38.81 -> 39.43 | Improved |
| model human and object features instead of object only (ablation) | VRDiff · HICO-DET UV | full mAP: 38.72 -> 39.43 | Improved |
| use human+object features instead of union region features (ablation) | VRDiff · HICO-DET UV | full mAP: 38.77 -> 39.43 | Improved |
| combine spatial and visual relation conditions instead of spatial only (ablation) | VRDiff · HICO-DET UV | full mAP: 38.01 -> 39.43 | Improved |
| combine spatial and visual relation conditions instead of visual only (ablation) | VRDiff · HICO-DET UV | full mAP: 37.80 -> 39.43 | Improved |
| use visual-only condition instead of spatial-only (ablation) | VRDiff · HICO-DET UV | full mAP: 38.01 -> 37.80 | Hurt |
| scale image encoder from ViT-B to ViT-L | VRDiff · HICO-DET Default | full mAP: 36.77 -> 41.40 | Improved |
Closest papers by meaning
- Human-Object Interaction Detection Collaborated with Large Relation-driven Diffusion Models (NeurIPS 2024 · similarity 0.94)
- An Image-like Diffusion Method for Human-Object Interaction Detection (CVPR 2025 · similarity 0.91)
- Toward Open-Set Human Object Interaction Detection (AAAI 2024 · similarity 0.89)
- Learning Human-Object Interaction as Groups (NeurIPS 2025 · similarity 0.89)
- Exploring Pose-Aware Human-Object Interaction via Hybrid Learning (CVPR 2024 · similarity 0.89)
- HORP: Human-Object Relation Priors Guided HOI Detection (CVPR 2025 · similarity 0.88)
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.
