Saved in:
| Main Authors: | , , , , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2602.18726 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866917304986501120 |
|---|---|
| author | Liang, Bo Gong, Chen Wang, Haobo Liu, Qirui Zhou, Rungui Shao, Fengzhi Wang, Yubo Gao, Wei Zhou, Kaichen Cui, Guolong Xu, Chenren |
| author_facet | Liang, Bo Gong, Chen Wang, Haobo Liu, Qirui Zhou, Rungui Shao, Fengzhi Wang, Yubo Gao, Wei Zhou, Kaichen Cui, Guolong Xu, Chenren |
| contents | Millimeter-wave Human Pose Estimation (mmWave HPE) promises privacy but suffers from poor generalization under distribution shifts. We demonstrate that brute-force data scaling is ineffective for out-of-distribution (OOD) robustness; efficiency and coverage are the true bottlenecks. To address this, we introduce WiCompass, a coverage-aware data-collection framework. WiCompass leverages large-scale motion-capture corpora to build a universal pose space ``oracle'' that quantifies dataset redundancy and identifies underrepresented motions. Guided by this oracle, WiCompass employs a closed-loop policy to prioritize collecting informative missing samples. Experiments show that WiCompass consistently improves OOD accuracy at matched budgets and exhibits superior scaling behavior compared to conventional collection strategies. By shifting focus from brute-force scaling to coverage-aware data acquisition, this work offers a practical path toward robust mmWave sensing. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2602_18726 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | WiCompass: Oracle-driven Data Scaling for mmWave Human Pose Estimation Liang, Bo Gong, Chen Wang, Haobo Liu, Qirui Zhou, Rungui Shao, Fengzhi Wang, Yubo Gao, Wei Zhou, Kaichen Cui, Guolong Xu, Chenren Computer Vision and Pattern Recognition Machine Learning Millimeter-wave Human Pose Estimation (mmWave HPE) promises privacy but suffers from poor generalization under distribution shifts. We demonstrate that brute-force data scaling is ineffective for out-of-distribution (OOD) robustness; efficiency and coverage are the true bottlenecks. To address this, we introduce WiCompass, a coverage-aware data-collection framework. WiCompass leverages large-scale motion-capture corpora to build a universal pose space ``oracle'' that quantifies dataset redundancy and identifies underrepresented motions. Guided by this oracle, WiCompass employs a closed-loop policy to prioritize collecting informative missing samples. Experiments show that WiCompass consistently improves OOD accuracy at matched budgets and exhibits superior scaling behavior compared to conventional collection strategies. By shifting focus from brute-force scaling to coverage-aware data acquisition, this work offers a practical path toward robust mmWave sensing. |
| title | WiCompass: Oracle-driven Data Scaling for mmWave Human Pose Estimation |
| topic | Computer Vision and Pattern Recognition Machine Learning |
| url | https://arxiv.org/abs/2602.18726 |