Research map / Video / Action recognition
Common Action Discovery and Localization in Unconstrained Videos
Jiong Yang, Junsong Yuan
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 |
|---|---|---|---|
| use the relaxed density (eta>0) subgraph selection instead of average-degree density (eta=0) (ablation) | subgraph selection for common action discovery · UCF Sports Plus (IoU 0.5) | average precision: 5.88% -> 50.29% | Improved |
| select proposals with the proposed method instead of using original proposals | C3D tube proposals · UCF Sports Plus (IoU 0.5) | average precision: 31.27% -> 50.29% | Improved |
| select proposals with the proposed method instead of using original proposals | C3D tube proposals · SVW Mini (IoU 0.25) | average precision: 25.60% -> 48.17% | Improved |
| replace graph cut selection with the proposed selection (ablation) | subgraph selection · UCF Sports Plus (IoU 0.5) | average precision: 31.28% -> 50.29% | Improved |
| replace DomSet clustering with the proposed selection (ablation) | subgraph selection · UCF Sports Plus (IoU 0.5) | average precision: 41.79% -> 50.29% | Improved |
| replace DomSet clustering with the proposed selection (ablation) | subgraph selection · SVW Mini (IoU 0.25) | average precision: 46.75% -> 48.17% | Improved |
| replace K-Means outlier removal with the proposed selection (ablation) | subgraph selection · UCF Sports Plus (IoU 0.5) | average precision: 32.24% -> 50.29% | Improved |
| apply the proposed selection before K-Means clustering of proposals | action clustering · UCF Sports Plus | clustering F-measure (average): 0.29 -> 0.37 | Improved |
| apply the proposed selection before K-Means clustering of proposals | action clustering · SVW Mini | clustering F-measure (average): 0.17 -> 0.32 | Improved |
| apply the proposed selection to larger sets with added outlier videos | input proposals · JHMDB + outlier videos (IoU 0.5) | mAP: 64.83% -> 69.00% | Improved |
| apply the proposed selection to larger sets with added outlier videos | input proposals · UCF101 24-class + outlier videos (IoU 0.25) | mAP: 26.07% -> 37.92% | Improved |
| use the proposed co-localization instead of the video object co-localization method [16] | common action co-localization · UCF Sports Plus (IoU 0.5, top-1, no outlier videos) | correct detection ratio: 31.85% -> 49.20% | Improved |
Closest papers by meaning
- Action Search: Spotting Actions in Videos and Its Application to Temporal Action Localization (ECCV 2018 · similarity 0.90)
- Few-Shot Transformation of Common Actions Into Time and Space (CVPR 2021 · similarity 0.90)
- Unsupervised Action Discovery and Localization in Videos (ICCV 2017 · similarity 0.90)
- Graph Convolutional Module for Temporal Action Localization in Videos (TPAMI 2021 · similarity 0.90)
- Few-Shot Common Action Localization via Cross-Attentional Fusion of Context and Temporal Dynamics (ICCV 2023 · similarity 0.89)
- Localizing the Common Action Among a Few Videos (ECCV 2020 · similarity 0.89)
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.
