Robust In-Bed Human Pose and Shape Estimation from Pressure Images with Clinical Awareness

作者：Chenfang Fang, Tingting Guo, Yi Ke, Mengting Niu, Xie F, Aiqin Chu, Yurong Qin, Changhai Ma, Quan Wan, Yunkang Zhang, Xiaohui Cai · 年份：2026 · DOI：10.1109/icassp55912.2026.11462963 · 研究领域：Human Pose and Action Recognition、Gait Recognition and Analysis、Balance, Gait, and Falls Prevention

Estimating the 3D pose and shape of in-bed patients is crucial for clinical applications such as pressure ulcer prevention and sleep quality monitoring. Pressure sensing offers a privacy-preserving and unobtrusive alternative to vision and wearable modalities, but progress is limited by the lack of realistic datasets and methods designed for complex clinical environments. We introduce a clinically grounded dataset collected in simulated intensive care settings, featuring diverse bed configurations and precise 3D annotations. We further propose a novel framework that improves robustness to pressure blur and environmental variation through blur-aware attention and bed-conditioned modulation. To align learning with clinical priorities, we design a risk-weighted evaluation metric that emphasizes regions at high risk of pressure ulcers. Together, these contributions enable more robust and clinically meaningful deployment, achieving state-of-the-art performance across all metrics with a 7.5% MPJPE reduction. Project page: https://github.com/cf-x/chips.