Research map / Detection and segmentation / Open-set and few-shot detection
Subjects and Their Objects: Localizing Interactees for a Person-Centric View of Importance
Chao-Yeh Chen, Kristen Lorraine Grauman
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 interaction-guided fine-tuned CNN features (ablation) | Ours-embedding · Interactee localization (COCO) | IOU: 0.1780 -> 0.1989 | Improved |
| add interaction-guided fine-tuned CNN features (ablation) | Ours-embedding · Interactee localization (PASCAL) | IOU: 0.1998 -> 0.2177 | Improved |
| add interaction-guided fine-tuned CNN features (ablation) | Ours-embedding · Interactee localization (SUN) | Position error: 0.2453 -> 0.2524 | Mixed |
| replace MDN with data-driven embedding regression | interactee predictor · Interactee localization (COCO) | IOU: 0.1153 -> 0.1989 | Improved |
| use embedding instead of MDN on tiny dataset | interactee predictor · Interactee localization (SUN) | Position error: 0.2356 -> 0.2524 | Hurt |
| predict interactee instead of Near Person heuristic | interactee predictor · Interactee localization (COCO) | Position error: 0.2909 -> 0.2256 | Improved |
| use MDN instead of Near Person heuristic | Ours-MDN · Interactee localization (COCO) | IOU: 0.1213 -> 0.1153 | Hurt |
| prune DPM detections outside the predicted interactee box | DPM detector · PASCAL (computer, reading materials) | detection accuracy | Improved |
| use predicted interactee to rank important objects instead of Importance method | object importance · COCO | Mention rate (%): 65.4 -> 70.5 | Improved |
| use ground truth interactee rather than predicted | object importance · COCO | Mention rate (%): 70.5 -> 78.4 | Improved |
| add interactee-conditioned retrieval instead of Global Matching | sentence retrieval · COCO | Combined BLEU: 16.01 -> 26.43 | Improved |
| add local Content Matching to Global Matching | sentence retrieval · COCO | Combined BLEU: 16.01 -> 15.20 | Hurt |
Closest papers by meaning
- Detecting and Recognizing Human-Object Interactions (CVPR 2018 · similarity 0.91)
- PoseSOR: Human Pose Can Guide Our Attention (ECCV 2024 · similarity 0.90)
- Learning Human-Object Interaction Detection Using Interaction Points (CVPR 2020 · similarity 0.90)
- Pose-Aware Multi-Level Feature Network for Human Object Interaction Detection (ICCV 2019 · similarity 0.89)
- Learning Human-Object Interaction as Groups (NeurIPS 2025 · similarity 0.87)
- Exploring Pose-Aware Human-Object Interaction via Hybrid Learning (CVPR 2024 · 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.
