Research map / Video / Action recognition
Self-Regulated Learning for Egocentric Video Activity Anticipation
Zhaobo Qi, Shuhui Wang, Chi Su, Li Su, Qingming Huang, Qi Chuan Tian
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 representation revision (contrastive loss) to baseline (ablation) | recursive GRU baseline · EPIC-Kitchens val, activity, tau_a=1s | top-5 accuracy: 29.18 -> 30.27 | Improved |
| add dynamic reattending to baseline (ablation) | recursive GRU baseline · EPIC-Kitchens val, activity, tau_a=1s | top-5 accuracy: 29.18 -> 30.71 | Improved |
| add semantic context exploration to baseline (ablation) | recursive GRU baseline · EPIC-Kitchens val, activity, tau_a=1s | top-5 accuracy: 29.18 -> 30.39 | Improved |
| add all three components (full SRL) to baseline (ablation) | recursive GRU baseline · EPIC-Kitchens val, activity, tau_a=1s | top-5 accuracy: 29.18 -> 31.68 | Improved |
| add representation revision to reattending plus semantic context (ablation) | SRL · EPIC-Kitchens val, activity, tau_a=1s | top-5 accuracy: 30.77 -> 31.68 | Improved |
| sample negatives from all videos instead of same video (ablation) | SRL contrastive loss · EPIC-Kitchens val, activity, tau_a=1s | top-5 accuracy: 30.15 -> 30.31 | No clear effect |
| vary number of negative samples N (32 to 256) (ablation) | SRL contrastive loss · EPIC-Kitchens val, activity, tau_a=1s | top-5 accuracy: 30.15 -> 30.19 | No clear effect |
| use GRU instead of average pooling for aggregation (ablation) | SRL · EPIC-Kitchens val, activity, tau_a=1s | top-5 accuracy: 31.03 -> 31.68 | Improved |
| use GRU instead of LSTM for aggregation (ablation) | SRL · EPIC-Kitchens val, activity, tau_a=1s | top-5 accuracy: 30.79 -> 31.68 | Improved |
| add OBJ features to RGB (ablation) | SRL · EPIC-Kitchens val, activity, tau_a=1s | top-5 accuracy: 31.68 -> 35.32 | Improved |
| replace RU(Late) with SRL(Late) at 0.25s | late-fusion model · EPIC-Kitchens val, activity, tau_a=0.25s | top-5 accuracy: 37.61 -> 40.21 | Improved |
| add representation revision to baseline (ablation) | recursive GRU baseline · 50 Salads (observed 20%, predict 10%) | accuracy: 22.96 -> 27.87 | Improved |
Closest papers by meaning
- A Hybrid Egocentric Activity Anticipation Framework via Memory-Augmented Recurrent and One-Shot Representation Forecasting (CVPR 2022 · similarity 0.91)
- LSTA: Long Short-Term Attention for Egocentric Action Recognition (CVPR 2019 · similarity 0.88)
- Test-Time Adaptation for Combating Missing Modalities in Egocentric Videos (ICLR 2025 · similarity 0.87)
- Action Scene Graphs for Long-Form Understanding of Egocentric Videos (CVPR 2024 · similarity 0.87)
- EgoVid-5M: A Large-Scale Video-Action Dataset for Egocentric Videos Generation (NeurIPS 2025 · similarity 0.86)
- Progress-Aware Online Action Segmentation for Egocentric Procedural Task Videos (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.
