Research map / Video / Action recognition
Learning Asynchronous and Sparse Human-Object Interaction in Videos
Romero Morais, Vuong Le, Svetha Venkatesh, Truyen Tran
Check 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 |
|---|---|---|---|
| Jointly discover segment boundaries and labels with sparse, asynchronous interaction modeling | ASSIGN · CAD-120, joint segmentation | F1@0.10: Relational BiRNN 79.2 ± 2.5 -> 88.0 ± 1.8; Independent BiRNN 70.2 ± 5.5 -> 88.0 ± 1.8 | Improved |
| Use ASSIGN for multi-entity interactions and segment-level decisions | Bimanual Actions models · Bimanual Actions, joint segmentation | F1@0.10: Relational BiRNN 77.7 ± 3.9 -> 84.0 ± 2.0 | Improved |
| Add spatial message passing (ablation) | ASSIGN · CAD-120, joint segmentation; sub-activity | F1@0.10: 74.5 -> 88.0 | Improved |
| Add spatial message passing (ablation) | ASSIGN · CAD-120, joint segmentation; object affordance | F1@0.50: 74.4 -> 82.4 | Improved |
| Add segmentation loss for joint segmentation and labeling (ablation) | ASSIGN · CAD-120, joint segmentation; sub-activity | F1@0.10: 84.3 -> 88.0 | Improved |
| Use sparse asynchronous rather than dense updates (ablation) | ASSIGN · CAD-120, joint segmentation; sub-activity | F1@0.50: 70.3 -> 73.8 | Improved |
| Use sparse asynchronous rather than dense updates (ablation) | ASSIGN · CAD-120, joint segmentation; object affordance | F1@0.50: 79.8 -> 82.4 | Improved |
| Add dense-model pre-training (ablation) | ASSIGN · CAD-120, joint segmentation; sub-activity | F1@0.50: 71.6 -> 73.8 | Improved |
| Add dense-model pre-training (ablation) | ASSIGN · CAD-120, joint segmentation; object affordance | F1@0.50: 78.9 -> 82.4 | Improved |
| Add anticipation loss (ablation) | ASSIGN · CAD-120, joint segmentation; sub-activity | F1@0.10: 86.2 -> 88.0 | Improved |
| Add anticipation loss (ablation) | ASSIGN · CAD-120, joint segmentation; object affordance | F1@0.50: 80.4 -> 82.4 | Improved |
Closest papers by meaning
- Attend and Interact: Higher-Order Object Interactions for Video Understanding (CVPR 2018 · similarity 0.88)
- Action Genome: Actions As Compositions of Spatio-Temporal Scene Graphs (CVPR 2020 · similarity 0.88)
- Hierarchical Video Prediction Using Relational Layouts for Human-Object Interactions (CVPR 2021 · similarity 0.88)
- Recurrent Space-time Graph Neural Networks (NeurIPS 2019 · similarity 0.88)
- Learning to Evolve Scenes: Reasoning about Human Activities with Scene Graphs (NeurIPS 2026 · similarity 0.87)
- SPARTAN: A Sparse Transformer World Model Attending to What Matters (NeurIPS 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.
