Research map / Detection and segmentation / Open-set and few-shot detection
Grounded Human-Object Interaction Hotspots From Video
Tushar Nagarajan, Christoph Feichtenhofer, Kristen Grauman
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 |
|---|---|---|---|
| increase backbone feature resolution to N=28 (ablation) | Ours (basic) · OPRA | AUC-J: 0.707 -> 0.766 | Improved |
| add L2 pooling (ablation) | Ours + resolution · OPRA | AUC-J: 0.766 -> 0.770 | Improved |
| add anticipation module (ablation) | Ours + resolution + L2 pool · OPRA | AUC-J: 0.770 -> 0.806 | Improved |
| add anticipation module (ablation) | Ours + resolution + L2 pool · EPIC | KLD: 1.361 -> 1.258 | Improved |
| add anticipation module (ablation) | Ours + resolution + L2 pool · EPIC | AUC-J: 0.727 -> 0.785 | Improved |
| add resolution (ablation) | Ours (basic) · EPIC | KLD: 1.342 -> 1.343 | No clear effect |
| add L2 pooling (ablation) | Ours + resolution · EPIC | AUC-J: 0.731 -> 0.727 | No clear effect |
| propagate gradients through anticipation to the input image rather than use the hypothesized output (ablation) | Ours · OPRA | AUC-J: 0.723 -> 0.806 | Improved |
| add anticipation module versus LSTM+Grad-CAM | LSTM action recognition · OPRA | KLD: 8.573 -> 1.427 | Improved |
| learn from weak action labels rather than saliency (SalGAN) | hotspot predictor · OPRA | KLD: 2.116 -> 1.427 | Improved |
| use strong keypoint supervision (Demo2Vec) instead of weak labels | Ours · OPRA | KLD: 1.427 -> 1.197 | Improved |
Closest papers by meaning
- Learning Affordance Grounding From Exocentric Images (CVPR 2022 · similarity 0.90)
- H2OFlow: Grounding Human-Object Affordances with 3D Generative Models and Dense Diffused Flows (ICLR 2026 · similarity 0.90)
- LOCATE: Localize and Transfer Object Parts for Weakly Supervised Affordance Grounding (CVPR 2023 · similarity 0.90)
- Weakly-Supervised Affordance Grounding Guided by Part-Level Semantic Priors (ICLR 2025 · similarity 0.89)
- IAAO: Interactive Affordance Learning for Articulated Objects in 3D Environments (CVPR 2025 · similarity 0.88)
- Weakly Supervised Multimodal Affordance Grounding for Egocentric Images (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.
