Research map / Video / Action recognition
Joint Prediction of Activity Labels and Starting Times in Untrimmed Videos
Tahmida Mahmud, Mahmudul Hasan, Amit K. Roy-Chowdhury
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 scene context (object features) (ablation) | Proposed network · MPII-Cooking | top-1 accuracy (%): 33.1 -> 80.1 | Improved |
| add scene context (object features) (ablation) | Proposed network · VIRAT Ground | top-1 accuracy (%): 61.0 -> 71.8 | Improved |
| add inter-activity time context branch (ablation) | Proposed network · MPII-Cooking | top-1 accuracy (%): 75.1 -> 80.1 | Improved |
| add inter-activity time context branch (ablation) | Proposed network · VIRAT Ground | top-1 accuracy (%): 69.2 -> 71.8 | Improved |
| add sequential activity and scene context to time prediction (ablation) | Proposed network · MPII-Cooking | RMSE (sec): 1.4872 -> 1.2454 | Improved |
| use LSTM sequence length 3 instead of 2 (ablation) | Proposed network · MPII-Cooking | top-1 accuracy (%): 78.8 -> 80.1 | Improved |
| use sequence length 3 instead of 9 (ablation) | Proposed network · MPII-Cooking | top-1 accuracy (%): 77.2 -> 80.1 | Improved |
| use sequence length 3 instead of 5 (ablation) | Proposed network · MPII-Cooking | top-1 accuracy (%): 79.2 -> 80.1 | Improved |
| use C3D features instead of MBH (ablation) | Proposed network · MPII-Cooking | top-1 accuracy (%): 80.1 -> 79.9 | No clear effect |
| use proposed prediction network instead of CNN + LSTM recognition | CNN + LSTM · MPII-Cooking | top-1/precision: 34.8 -> 70.7 | Improved |
| use proposed network instead of graphical model | Graphical Model · VIRAT Ground | top-1 accuracy (%): 68.5 -> 71.8 | Improved |
| use proposed network instead of statistical LGCP model for time | Statistical Model · MPII-Cooking | RMSE (sec): 3.9431 -> 1.2454 | Improved |
Closest papers by meaning
- Multi-Label Activity Recognition Using Activity-Specific Features and Activity Correlations (CVPR 2021 · similarity 0.89)
- Weakly-Guided Self-Supervised Pretraining for Temporal Activity Detection (AAAI 2023 · similarity 0.88)
- Fast Temporal Activity Proposals for Efficient Detection of Human Actions in Untrimmed Videos (CVPR 2016 · similarity 0.88)
- METAL: Minimum Effort Temporal Activity Localization in Untrimmed Videos (CVPR 2020 · similarity 0.88)
- Learning Activity Progression in LSTMs for Activity Detection and Early Detection (CVPR 2016 · similarity 0.87)
- HARDVS: Revisiting Human Activity Recognition with Dynamic Vision Sensors (AAAI 2024 · 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.
