Research map / Detection and segmentation / Weakly supervised detection and segmentation
Detector-Free Weakly Supervised Group Activity Recognition
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 partial context embedding to a global-scene base model (ablation) | Detector-Free WSGAR · NBA | MCA: 58.4 -> 64.1 | Helped |
| Add partial context aggregation to PCE (ablation) | Detector-Free WSGAR · NBA | MCA: 64.1 -> 73.6 | Helped |
| Add motion feature modules (ablation) | Detector-Free WSGAR · NBA | MCA: 73.6 -> 75.8 | Helped |
| Aggregate temporal embeddings with 1D convolution before aggregating tokens with MHSA, rather than reversing the order (ablation) | Two-stage token aggregation · NBA | MCA: 72.4 -> 73.6 | Helped |
| Aggregate temporal embeddings with 1D convolution before aggregating tokens with MHSA, rather than reversing the order (ablation) | Two-stage token aggregation · NBA | MPCA: +2.0%p of MPCA | Helped |
| Use multiple learnable tokens rather than a single token (ablation) | Partial context embedding · NBA | MCA | Helped |
| Insert motion modules after residual blocks 4 and 5 rather than after block 2 (ablation) | Motion-augmented backbone · NBA | MCA | Helped |
| Use wider-receptive-field temporal convolution layers (ablation) | Partial context aggregation · NBA | MCA | Helped |
| Use three temporal convolution layers with kernel size 5 instead of three with kernel size 3 (ablation) | Partial context aggregation · NBA | MCA: 72.7 -> 73.6 | Helped |
Closest papers by meaning
- Learning Visual Context for Group Activity Recognition (AAAI 2021 · similarity 0.90)
- Actor-Transformers for Group Activity Recognition (CVPR 2020 · similarity 0.90)
- A Hierarchical Deep Temporal Model for Group Activity Recognition (CVPR 2016 · similarity 0.89)
- Activity Driven Weakly Supervised Object Detection (CVPR 2019 · similarity 0.88)
- Learning Actor Relation Graphs for Group Activity Recognition (CVPR 2019 · similarity 0.88)
- Bi-Causal: Group Activity Recognition via Bidirectional Causality (CVPR 2024 · 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.