Research map / Detection and segmentation / Open-set and few-shot detection
Learning Transferable Human-Object Interaction Detector With Natural Language Supervision
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Use joint visual-and-text matching with pretrained CLIP features | THID · HICO-DET unseen | mAP: 13.16 -> 15.53 | Helped |
| Add HOI sequence parsing so tokens account for preceding detections (ablation) | THID · SWIG-HOI unseen | mAP: 6.27 -> 10.04 | Helped |
| Add HOI sequence parsing so tokens account for preceding detections (ablation) | THID · SWIG-HOI non-rare | mAP: 12.26 -> 17.67 | Helped |
| Replace sequential parsing with SetParser (ablation) | THID · SWIG-HOI non-rare | mAP: 17.67 -> 16.93 | Hurt |
| Replace sequential parsing with SetParser (ablation) | THID · SWIG-HOI unseen | mAP: 10.04 -> 7.07 | Hurt |
| Include the pretrained [CLS] token in the sequence parser (ablation) | THID · SWIG-HOI unseen | mAP: 8.43 -> 10.04 | Helped |
| Use learnable prefix and conjunction tokens instead of a fixed interaction sentence template (ablation) | THID · SWIG-HOI unseen | mAP: 6.96 -> 10.04 | Helped |
| Add the text-to-visual classification loss (ablation) | THID · SWIG-HOI unseen | mAP: 9.30 -> 10.04 | Helped |
| Use hold-out cross-entropy loss rather than multi-label soft margin loss (ablation) | THID · SWIG-HOI unseen | mAP: 6.02 -> 10.04 | Helped |
| Append action definitions to the fixed text template (ablation) | THID · SWIG-HOI unseen | mAP: 6.96 -> 8.22 | Helped |
| Append object definitions to the fixed text template (ablation) | THID · SWIG-HOI unseen | mAP: 6.96 -> 7.70 | Helped |
Closest papers by meaning
- LINK: Learning Instance-level Knowledge from Vision-Language Models for Human-Object Interaction Detection (ICLR 2026 · similarity 0.94)
- Toward Open-Set Human Object Interaction Detection (AAAI 2024 · similarity 0.92)
- Exploring Pose-Aware Human-Object Interaction via Hybrid Learning (CVPR 2024 · similarity 0.91)
- Discovering Syntactic Interaction Clues for Human-Object Interaction Detection (CVPR 2024 · similarity 0.91)
- Exploring the Potential of Large Foundation Models for Open-Vocabulary HOI Detection (CVPR 2024 · similarity 0.91)
- Open-World Human-Object Interaction Detection via Multi-modal Prompts (CVPR 2024 · similarity 0.90)
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.