Research map / Detection and segmentation / Weakly supervised detection and segmentation
Weakly Supervised Human-Object Interaction Detection in Video via Contrastive Spatiotemporal Regions
Shuang Li, Yilun Du, Antonio Torralba, Josef Sivic, Bryan Russell
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 sparsity loss (ablation) | concatenation baseline · V-HICO test (seen classes) | Phrase mAP (ko): 41.86 -> 50.79 | Improved |
| add sparsity loss (ablation) | concatenation baseline · V-HICO test (seen classes) | Relation mAP (ko): 11.34 -> 16.23 | Improved |
| add human ROI-pooled feature to object feature (ablation) | (cat)+Spa · V-HICO test (seen classes) | Phrase mAP (ko): 50.79 -> 55.60 | Improved |
| add human ROI-pooled feature to object feature (ablation) | (cat)+Spa · V-HICO test (seen classes) | Relation mAP (ko): 16.23 -> 15.91 | No clear effect |
| add temporal contrastive loss (ablation) | (cat)+Spa+Hum · V-HICO test (seen classes) | Relation mAP (ko): 15.91 -> 16.94 | Improved |
| add temporal contrastive loss (ablation) | (cat)+Spa+Hum · V-HICO test (seen classes) | Phrase mAP (ko): 55.60 -> 54.42 | Hurt |
| add contextual frame feature (ablation) | (cat)+Spa+Hum+Tem · V-HICO test (seen classes) | Relation mAP (ko): 16.94 -> 18.56 | Improved |
| add contextual frame feature (ablation) | (cat)+Spa+Hum+Tem · V-HICO test (seen classes) | Relation Recall@1: 25.61 -> 30.12 | Improved |
| concatenate instead of add features (ablation) | Baseline · V-HICO test (seen classes) | Phrase mAP (ko): 40.59 -> 41.86 | No clear effect |
| use learned full model instead of human spatial prior alone (ablation) | human-prior baseline · V-HICO test (seen classes) | Phrase mAP (ko): 46.67 -> 55.90 | Improved |
| full method vs prior method on unseen object classes | Ours vs Peyre2017 · V-HICO unseen test (51 unseen object classes) | Phrase mAP (ko): 38.19 -> 67.21 | Improved |
Closest papers by meaning
- Track and Transfer: Watching Videos to Simulate Strong Human Supervision for Weakly-Supervised Object Detection (CVPR 2016 · similarity 0.89)
- Weakly-supervised HOI Detection via Prior-guided Bi-level Representation Learning (ICLR 2023 · similarity 0.89)
- Learning Human-Object Interaction Detection Using Interaction Points (CVPR 2020 · similarity 0.89)
- Weakly-Supervised Action Localization With Background Modeling (ICCV 2019 · similarity 0.88)
- Activity Driven Weakly Supervised Object Detection (CVPR 2019 · similarity 0.88)
- Interacted Object Grounding in Spatio-Temporal Human-Object Interactions (AAAI 2025 · similarity 0.86)
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.
