ChairPose: Pressure-based Chair Morphology Grounded Sitting Pose Estimation through Simulation-As...
summary
A system estimates full-body seated pose from a thin chair-agnostic pressure mattress plus a single 3D chair scan, using a two-stage generative model with a learned motion codebook and an autoregressive pressure-to-pose network. A Blender-based pressure simulator provides physics-driven data augmentation to generalize across chairs and users. Evaluated on 8 users and 4 chairs it reports a mean per joint position error of 89.4 mm when both chair and user are unseen and runs in real time. pressure mattress + 3D scan physics-driven data augmentation
AI-written summary · about
chapters (7)
- 0:00Introduction to ChairPose
- 1:53Pressure sensors for pose estimation
- 3:37Three pillars of the method
- 6:49Synthetic data generation
- 7:53Evaluation and results
- 9:40Limitations and future work
- 10:00Q&A session
related talks
17mEMPress: Practical Hand Gesture Classification with Wrist-Mounted EMG and Pressure Sensing
06mBodyCompass: Monitoring Sleep Posture with Wireless Signals
06mShadowSense: Detecting Human Touch in a Social Robot Using Shadow Image Classification
10mDexmo: An Inexpensive and Lightweight Mechanical Exoskeleton for Motion Capture and Force ...
21mPmomo: Projection Mapping on Movable 3D Object
23mMetaArms: Body Remapping Using Feet-Controlled Artificial Arms
18mSWISH: A Shifting-Weight Interface of Simulated Hydrodynamics for Haptic Perception of Virtual ...
08mCapturing Tactile Properties of Real Surfaces for Haptic Reproduction