Research map / Video / Video-language understanding
Learning Consistent Temporal Grounding between Related Tasks in Sports Coaching
Arushi Rai, Adriana Kovashka
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 dual-pathway self-consistency training (full method) (ablation) | PerceptionLM-8B finetuned · Exact | accuracy: 84.5 -> 87.5 | Improved |
| add dual-pathway self-consistency training (full method) (ablation) | PerceptionLM-8B finetuned · FitnessQA | accuracy: 74.1 -> 88.2 | Improved |
| add dual-pathway self-consistency training (full method) (ablation) | PerceptionLM-8B finetuned · ExpertAF | BERTScore: 37.5 -> 38.4 | Improved |
| redistribute attention proportionally over ground-truth keyframes (ablation) | PerceptionLM-8B (manual head) · EgoExo4D split of VidDiffBench | accuracy: 65.8 -> 71.8 | Improved |
| select heads by visual non-sink ratio instead of visual attention sum (ablation) | keyframe attention redistribution · EgoExo4D split of VidDiffBench | accuracy: 69.2 -> 73.4 | Improved |
| match unmatched heads with Hungarian algorithm instead of random assignment (ablation) | head matching · Exact | accuracy: 83.0 -> 87.5 | Improved |
| match unmatched heads with Hungarian algorithm instead of discarding them (ablation) | head matching · FitnessQA | accuracy: 81.3 -> 88.2 | Improved |
| share the same layer across pathways instead of flexible verification layer (ablation) | L_consistency · Exact | accuracy: 83.0 -> 87.5 | Improved |
| use verification task instead of summarization as the related task (ablation) | L_consistency · Exact | accuracy: 69.5 -> 87.5 | Improved |
| use K=5 heads instead of larger K (ablation) | head selection · Exact | accuracy: 63.5 -> 87.5 | Improved |
| add L_consistency to L_ce and L_entropy (ablation) | self-consistency training · Exact | accuracy: 77.5 -> 87.5 | Improved |
| add L_entropy to L_ce and L_consistency (ablation) | self-consistency training · ExpertAF | BERTScore: 36.8 -> 38.3 | Improved |
Closest papers by meaning
- ExpertAF: Expert Actionable Feedback from Video (CVPR 2025 · similarity 0.87)
- Learning Skill-Attributes for Transferable Assessment in Video (NeurIPS 2025 · similarity 0.86)
- SoccerMaster: A Vision Foundation Model for Soccer Understanding (CVPR 2026 · similarity 0.85)
- SportR: A Benchmark for Multimodal Large Language Model Reasoning in Sports (ICLR 2026 · similarity 0.84)
- ViSTec: Video Modeling for Sports Technique Recognition and Tactical Analysis (AAAI 2024 · similarity 0.84)
- From 3D Pose to Prose: Biomechanics-Grounded Vision-Language Coaching (CVPR 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.
