Research map / Video / Action recognition
Temporal Dynamic Graph LSTM for Action-Driven Video Object Detection
Yuan Yuan, Xiaodan Liang, Xiaolong Wang, Dit-Yan Yeung, Abhinav Gupta
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 dynamic temporal graph between region proposals (ablation) | TD-Graph LSTM · Charades (action-driven weak supervision) | classification mAP: 17.13 -> 19.52 | Improved |
| add dynamic temporal graph between region proposals (ablation) | TD-Graph LSTM · Charades (action-driven weak supervision) | detection mAP: 1.55 -> 1.98 | Improved |
| add LSTM units over the temporal graph (ablation) | TD-Graph LSTM · Charades (action-driven weak supervision) | classification mAP: 15.89 -> 19.52 | Improved |
| add LSTM units over the temporal graph (ablation) | TD-Graph LSTM · Charades (action-driven weak supervision) | detection mAP: 1.16 -> 1.98 | Improved |
| use dynamic graph instead of static graph with adaptive edge weights (ablation) | TD-Graph LSTM · Charades (action-driven weak supervision) | classification mAP: 17.97 -> 19.52 | Improved |
| use dynamic graph instead of static graph with adaptive edge weights (ablation) | TD-Graph LSTM · Charades (action-driven weak supervision) | detection mAP: 1.89 -> 1.98 | Improved |
| use dynamic graph instead of mean graph (uniform edge weights) (ablation) | TD-Graph LSTM · Charades (action-driven weak supervision) | classification mAP: 16.92 -> 19.52 | Improved |
| use dynamic graph instead of mean graph (uniform edge weights) (ablation) | TD-Graph LSTM · Charades (action-driven weak supervision) | detection mAP: 1.41 -> 1.98 | Improved |
| use a fully-connected mean graph instead of no graph (ablation) | TD-Graph LSTM · Charades (action-driven weak supervision) | detection mAP: 1.55 -> 1.41 | Hurt |
| use TD-Graph LSTM instead of ContextLocNet | weakly-supervised detector · Charades (action-driven weak supervision) | classification mAP: 16.47 -> 19.52 | Improved |
| use TD-Graph LSTM instead of WSDDN | weakly-supervised detector · Charades (action-driven weak supervision) | classification mAP: 15.67 -> 19.52 | Improved |
| use TD-Graph LSTM instead of ContextLocNet on small objects | weakly-supervised detector · Charades (action-driven weak supervision), pillow class | classification AP: 3.51 -> 17.64 | Improved |
Closest papers by meaning
- Activity Driven Weakly Supervised Object Detection (CVPR 2019 · similarity 0.89)
- G-TAD: Sub-Graph Localization for Temporal Action Detection (CVPR 2020 · similarity 0.88)
- Unified Graph Structured Models for Video Understanding (ICCV 2021 · similarity 0.88)
- Lattice Long Short-Term Memory for Human Action Recognition (ICCV 2017 · similarity 0.88)
- Large-Scale Weakly-Supervised Pre-Training for Video Action Recognition (CVPR 2019 · similarity 0.88)
- Uncertainty-Aware Weakly Supervised Action Detection from Untrimmed Videos (ECCV 2020 · 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.
