Research map / Detection and segmentation / Open-set and few-shot detection
Category-Aware Transformer Network for Better Human-Object Interaction Detection
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 |
|---|---|---|---|
| Initialize the decoder queries with category-aware priors instead of zeros (ablation) | QPIC on HICO-DET · HICO-DET Default Full | mAP: 29.07 -> 30.82 | Helped |
| Add modified matching cost to CAQ (ablation) | CATN on HICO-DET · HICO-DET Default Full | mAP: 30.82 -> 31.03 | Helped |
| Add CLAM to CAQ and modified matching cost (ablation) | CATN on HICO-DET · HICO-DET Default Full | mAP: 31.03 -> 31.62 | Helped |
| Replace zero query initialization with ground-truth category priors | QPIC on HICO-DET · HICO-DET Default Full (oracle) | mAP: 29.07 -> 37.17 | Helped |
| Use category-aware semantic initialization rather than zero or random initialization (ablation) | QPIC on HICO-DET · HICO-DET Default Full | mAP: 29.07 -> 31.03; 29.70; 29.60 | Helped |
| Use BERT rather than fastText category embeddings (ablation) | CATN on HICO-DET · HICO-DET Default Full | mAP: 31.62 -> 31.86 | Helped |
| Use CLIP rather than fastText category embeddings (ablation) | CATN on HICO-DET · HICO-DET Default Full | mAP: 31.62 -> 31.71 | No clear effect |
| Apply category priors as query initialization rather than as complementary prediction-head features (ablation) | CATN on HICO-DET · HICO-DET Default Full | mAP | Helped |
| Apply HOI-NMS to CATN predictions (ablation) | CATN on HICO-DET · HICO-DET Default Full | mAP: 31.62 -> 32.38 | Helped |
| Apply HOI-SoftNMS to CATN predictions (ablation) | CATN on HICO-DET · HICO-DET Default Full | mAP: 31.62 -> 32.40 | Helped |
| Use Yolov5m instead of Faster R-CNN and run the detector in parallel (ablation) | CATN on HICO-DET · HICO-DET Default Full | inference time: 81.3 -> 46.1 | Helped |
Closest papers by meaning
- QueryCraft: Transformer-Guided Query Initialization for Enhanced Human-Object Interaction Detection (AAAI 2026 · similarity 0.92)
- Exploring Structure-Aware Transformer Over Interaction Proposals for Human-Object Interaction Detection (CVPR 2022 · similarity 0.91)
- End-to-End Human Object Interaction Detection With HOI Transformer (CVPR 2021 · similarity 0.91)
- HOTR: End-to-End Human-Object Interaction Detection With Transformers (CVPR 2021 · similarity 0.90)
- Orchestrating the Symphony of Prompt Distribution Learning for Human-Object Interaction Detection (AAAI 2025 · similarity 0.89)
- Dual-Prior Augmented Decoding Network for Long Tail Distribution in HOI Detection (AAAI 2024 · 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.