Tracking human sleeping postures over time provides critical information to biomedical research including studies on sleeping behaviors and bedsore prevention. In this work, we introduce a vision-based tracking system for pervasive yet unobtrusive long-term monitoring of in-bed postures in different environments. In this project, we estimation lying posture with low granularity into left, right and supine position. A latent parameter is introduced in the context of narrow field of view in our application during posture recognition, which improves the classification accuracy and also acts in the role of occupation detection.
Selected publication:
- “A Vision-Based System for In-Bed Posture Tracking,” ICCV/ACVR’17. [Code]
Related download:
- Mannequin dataset (downsampled): manneSep2