Learning Consistent Temporal Grounding between Related Tasks in Sports Coaching
Fuente:
arXiv
Enregistré dans:
| Auteurs principaux: | , |
|---|---|
| Format: | Preprint |
| Publié: |
2026
|
| Sujets: | |
| Accès en ligne: | |
| Tags: |
Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
|
| _version_ | 1866911528842690560 |
|---|---|
| author | Rai, Arushi Kovashka, Adriana |
| author_facet | Rai, Arushi Kovashka, Adriana |
| contents | Video-LLMs often attend to irrelevant frames, which is especially detrimental for sports coaching tasks requiring precise temporal grounding. Yet obtaining frame-level supervision is challenging: expensive to collect from humans and unreliable from other models. We improve temporal grounding without additional annotations by exploiting the observation that related tasks, such as generation and verification, must attend to the same frames. We enforce this via a self-consistency objective over select visual attention maps of tightly-related tasks. Using VidDiffBench, which provides ground-truth keyframe annotations, we first validate that attention misallocation is a significant bottleneck. We then show that training with our objective yields gains of +3.0%, +14.1% accuracy and +0.9 BERTScore over supervised finetuning across three sports coaching tasks: Exact, FitnessQA, and ExpertAF, even surpassing closed-source models. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2603_18453 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Learning Consistent Temporal Grounding between Related Tasks in Sports Coaching Rai, Arushi Kovashka, Adriana Computer Vision and Pattern Recognition Video-LLMs often attend to irrelevant frames, which is especially detrimental for sports coaching tasks requiring precise temporal grounding. Yet obtaining frame-level supervision is challenging: expensive to collect from humans and unreliable from other models. We improve temporal grounding without additional annotations by exploiting the observation that related tasks, such as generation and verification, must attend to the same frames. We enforce this via a self-consistency objective over select visual attention maps of tightly-related tasks. Using VidDiffBench, which provides ground-truth keyframe annotations, we first validate that attention misallocation is a significant bottleneck. We then show that training with our objective yields gains of +3.0%, +14.1% accuracy and +0.9 BERTScore over supervised finetuning across three sports coaching tasks: Exact, FitnessQA, and ExpertAF, even surpassing closed-source models. |
| title | Learning Consistent Temporal Grounding between Related Tasks in Sports Coaching |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2603.18453 |