Research map / Video / Video-language understanding
Weakly Supervised Temporal Sentence Grounding With Uncertainty-Guided Self-Training
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 uncertainty-guided teacher-student self-training (Bayesian teacher weighting) | CPL backbone · Charades-STA test | mIoU: 43.48 -> 45.20 | Helped |
| add uncertainty-guided self-training | CPL backbone · ActivityNet Captions val 2 | mIoU: 35.55 -> 41.02 | Helped |
| apply standard Mean Teacher self-training | CPL backbone · Charades-STA test | mIoU: 43.48 -> 39.31 | Hurt |
| apply pseudo-label self-training (ablation) | CPL backbone · Charades-STA test | mIoU: 43.48 -> 40.24 | Hurt |
| add Bayesian uncertainty weighting to pseudo-label self-training (ablation) | Pseudo label · Charades-STA test | mIoU: 40.24 -> 42.75 | Helped |
| add Bayesian uncertainty weighting to Mean Teacher (ablation) | Mean Teacher · Charades-STA test | mIoU: 39.31 -> 40.43 | Helped |
| remove Bayesian uncertainty weighting from full method (ablation) | Ours full · Charades-STA test | mIoU: 42.88 -> 45.20 when uncertainty added | Helped |
| add reconstruction ranking and reconstruction losses (ablation) | Ours full · Charades-STA test | mIoU: 43.39 -> 45.20 | Helped |
| add video temporal scaling/shifting augmentation (ablation) | CPL with self-training · Charades-STA test | IoU@0.3: 66.40 -> 68.02 | Helped |
| add SRL sentence query decomposition augmentation (ablation) | CPL with self-training · ActivityNet Captions val 2 | IoU@0.3: 53.99 -> 56.96 | Helped |
| apply method to CNM and WSTAN backbones | CNM, WSTAN · Charades-STA and ActivityNet Captions | IoU@0.3: Positive improvement reported for all three backbones | Helped |
| apply self-training method | CPL backbone · Charades-STA test Recall@5 | IoU@0.3 (Recall@5): CPL + aug 84.71/85.40 vs CPL+Ours 84.86 at IoU@0.3 (Recall@5) | Hurt |
Closest papers by meaning
- Weakly Supervised Temporal Sentence Grounding With Gaussian-Based Contrastive Proposal Learning (CVPR 2022 · similarity 0.91)
- Multi-Pair Temporal Sentence Grounding via Multi-Thread Knowledge Transfer Network (AAAI 2025 · similarity 0.88)
- Siamese Learning with Joint Alignment and Regression for Weakly-Supervised Video Paragraph Grounding (CVPR 2024 · similarity 0.88)
- Contextual Self-paced Learning for Weakly Supervised Spatio-Temporal Video Grounding (ICLR 2025 · similarity 0.87)
- Learning Multi-Scale Video-Text Correspondence for Weakly Supervised Temporal Article Gronding (AAAI 2024 · similarity 0.86)
- Seq2Time: Sequential Knowledge Transfer for Video LLM Temporal Grounding (CVPR 2025 · 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.