Research map / Video / Action recognition
Panoramic Human Activity Recognition
Ruize Han, Haomin Yan, Jiacheng Li, Songmiao Wang, Wei Feng, Song Wang
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 original individual feature f to node representation (ablation) | Ours · JRDB-PAR | overall F1: 25.1 -> 35.6 | Improved |
| add surrounding-aware feature to node representation (ablation) | Ours · JRDB-PAR | overall F1: 21.9 -> 35.6 | Improved |
| use area-normalized distance instead of Euclidean distance (ablation) | Ours · JRDB-PAR | overall F1: 26.0 -> 35.6 | Improved |
| add distance-aware matrix D-breve to relation matrix (ablation) | Ours · JRDB-PAR | overall F1: 19.4 -> 35.6 | Improved |
| add affinity edge matrix E to relation matrix (ablation) | Ours · JRDB-PAR | overall F1: 28.4 -> 35.6 | Improved |
| use AiO aggregation instead of max-pooling (ablation) | Ours · JRDB-PAR | overall F1: 21.6 -> 35.6 | Improved |
| add global node feature to individual nodes (G.2I.) (ablation) | Ours · JRDB-PAR | individual action F1: 25.3 -> 43.4 | Improved |
| add global node feature to group nodes (G.2P.) (ablation) | Ours · JRDB-PAR | social activity F1: 10.9 -> 24.8 | Improved |
| add global node feature to individual nodes (G.2I.) (ablation) | Ours · JRDB-PAR | global activity F1: 55.2 -> 38.8 | Hurt |
| provide ground-truth group detection as input | Ours · JRDB-PAR | overall F1: 35.6 -> 42.4 | Improved |
| provide ground-truth group detection as input | JRDB-Base · JRDB-PAR | overall F1: 27.2 -> 37.0 | Improved |
| provide ground-truth group detection as input | Ours · JRDB-PAR | social activity F1: 24.8 -> 50.1 | Improved |
Closest papers by meaning
- Skeleton-based Group Activity Recognition via Spatial-Temporal Panoramic Graph (ECCV 2024 · similarity 0.88)
- Joint Learning of Social Groups, Individuals Action and Sub-group Activities in Videos (ECCV 2020 · similarity 0.87)
- Hunting Group Clues with Transformers for Social Group Activity Recognition (ECCV 2022 · similarity 0.86)
- Social Scene Understanding: End-To-End Multi-Person Action Localization and Collective Activity Recognition (CVPR 2017 · similarity 0.86)
- A Hierarchical Deep Temporal Model for Group Activity Recognition (CVPR 2016 · similarity 0.85)
- HARDVS: Revisiting Human Activity Recognition with Dynamic Vision Sensors (AAAI 2024 · similarity 0.84)
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.
