Research map / Label-efficient and robust learning / Semi-supervised learning
Self-Supervised Learning for Semi-Supervised Temporal Action Proposal
Xiang Wang, Shiwei Zhang, Zhiwu Qing, Yuanjie Shao, Changxin Gao, Nong Sang
Check 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 SSTAP rather than Ji et al.'s semi-supervised baseline | BMN-based temporal action proposal · THUMOS14 test, 10% labels | AR@100: 37.43 -> 40.92 | Improved |
| Use SSTAP rather than the BMN baseline | BMN-based temporal action proposal · ActivityNet v1.3 validation, 60% labels | AR@100: 74.42 -> 75.20 | Improved |
| Add temporal feature flip (ablation) | SSTAP · THUMOS14 test, 10% labels | AR@100: 40.52 -> 40.92 | Improved |
| Add masked feature reconstruction to the variant without flip and clip-order prediction (ablation) | SSTAP · THUMOS14 test, 10% labels | AR@100: 38.75 -> 39.24 | Improved |
| Add clip-order prediction to the variant without flip or reconstruction (ablation) | SSTAP · THUMOS14 test, 10% labels | AR@100: 39.24 -> 40.52 | Improved |
| Add temporal feature shift to the variant without reconstruction or clip-order prediction (ablation) | SSTAP · THUMOS14 test, 10% labels | AR@100: 37.57 -> 40.92 | Improved |
| Add both self-supervised tasks (ablation) | SSTAP · THUMOS14 test, 60% labels | AR@100: 46.89 -> 48.02 | Improved |
| Add temporal feature flip (ablation) | SSTAP · THUMOS14 test, 60% labels | AR@100: 48.00 -> 48.02 | No clear effect |
| Add clip-order prediction and masked feature reconstruction (ablation) | BMN · THUMOS14 test, 10% labels | AR@100: 31.11 -> 36.13 | Improved |
| Add clip-order prediction and masked feature reconstruction (ablation) | BMN · THUMOS14 test, 60% labels | AR@100: 42.11 -> 47.00 | Improved |
Closest papers by meaning
- Learning Temporal Action Proposals With Fewer Labels (ICCV 2019 · similarity 0.90)
- Semi-Supervised Temporal Action Detection with Proposal-Free Masking (ECCV 2022 · similarity 0.87)
- A Perceptual Prediction Framework for Self Supervised Event Segmentation (CVPR 2019 · similarity 0.86)
- Action Segmentation With Joint Self-Supervised Temporal Domain Adaptation (CVPR 2020 · similarity 0.86)
- No More Shortcuts: Realizing the Potential of Temporal Self-Supervision (AAAI 2024 · similarity 0.84)
- HyrCap: Hybrid Rank-Calibration of Action Proposals for Temporal Event Understanding (NeurIPS 2026 · similarity 0.84)
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.
