HRTR: A Single-Stage Transformer for Fine-Grained Sub-Second Action Segmentation in Stroke Rehabilitation
Halil Ismail Helvaci Justin Philip Huber Jihye Bae Sen-ching Samson Cheung · 2025
Read HRTR: A Single-Stage Transformer for Fine-Grained Sub-Second Action Segmentation in Stroke Rehabilitation by Halil Ismail Helvaci Justin Philip Huber Jihye Bae Sen-ching Samson Cheung (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2506-02472.txt.
Description
Halil Ismail Helvaci and three co-authors build a single-stage transformer that slices rehabilitation exercises into sub-second actions, in this 2025 paper. After stroke, therapists track fine movement steps to judge progress. The steps are brief, often under a second, and finely distinguished, which defeats pipelines that first propose segments and then polish them.
Their High Resolution Temporal Transformer localizes and labels the actions in one pass, with no post-processing. On two stroke datasets, video and wearable-sensor based, it reached edit scores of about seventy, and on a general cooking-activity dataset it scored 88.4, beating prior state-of-the-art systems throughout. The authors present the model as a step toward automatic, fine-grained monitoring of rehabilitation at home and in clinics, where precise timing of small movements carries the clinical signal.
Catalog record
| CREATOR | Halil Ismail Helvaci Justin Philip Huber Jihye Bae Sen-ching Samson Cheung |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 23 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2506-02472