Research map / Detection and segmentation / Open-set and few-shot detection
UnionDet: Union-Level Detector Towards Real-Time Human-Object Interaction Detection
Bumsoo Kim, Taeho Choi, Jaewoo Kang, Hyunwoo J. Kim
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 the Union Branch to the Instance Branch (ablation) | UnionDet · V-COCO test (Scenario 1) | AP role: 38.4 -> 47.5 | Improved |
| add the Union Branch to the Instance Branch (ablation) | UnionDet · V-COCO test (Scenario 2) | AP role: 51.0 -> 56.2 | Improved |
| add target object classification loss (ablation) | UnionDet · V-COCO test (Scenario 1) | AP role: 44.8 -> 47.5 | Improved |
| add target object classification loss (ablation) | UnionDet · V-COCO test (Scenario 2) | AP role: 53.5 -> 56.2 | Improved |
| use union-instance matching score instead of standard IoU (ablation) | UnionDet · V-COCO test (Scenario 1) | AP role: 45.0 -> 47.5 | Improved |
| use union-instance matching score instead of standard IoU (ablation) | UnionDet · V-COCO test (Scenario 2) | AP role: 53.6 -> 56.2 | Improved |
| use Foreground Focal Loss instead of ordinary Focal Loss (ablation) | UnionDet · V-COCO test (Scenario 1) | AP role: 46.9 -> 47.5 | Improved |
| use Foreground Focal Loss instead of ordinary Focal Loss (ablation) | UnionDet · V-COCO test (Scenario 2) | AP role: 55.6 -> 56.2 | Improved |
| one-stage union detector instead of pairwise second-stage inference | UnionDet · V-COCO test | additional inference time (ms): 40ms ~ 130ms -> 9.06 ms | Improved |
| compare UnionDet with DCA | UnionDet · HICO-DET Default Full | mAP: 16.25 -> 17.58 | Improved |
| compare UnionDet with DCA | UnionDet · HICO-DET Known Object Full | mAP: 17.73 -> 19.76 | Improved |
| compare UnionDet with models using external pose features (PMFNet) | UnionDet · V-COCO test | AP role: 52.0 -> 47.5 | Hurt |
Closest papers by meaning
- Learning Human-Object Interaction Detection Using Interaction Points (CVPR 2020 · similarity 0.90)
- Human-Object Interaction Detection via Disentangled Transformer (CVPR 2022 · similarity 0.89)
- DIRV: Dense Interaction Region Voting for End-to-End Human-Object Interaction Detection (AAAI 2021 · similarity 0.89)
- Exploring Pose-Aware Human-Object Interaction via Hybrid Learning (CVPR 2024 · similarity 0.88)
- Exploring Self- and Cross-Triplet Correlations for Human-Object Interaction Detection (AAAI 2024 · similarity 0.87)
- HORP: Human-Object Relation Priors Guided HOI Detection (CVPR 2025 · similarity 0.87)
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.
