TouchPose: Hand Pose Prediction, Depth Estimation, and Touch Classification from Capacitive Images
summary
A multi-task convolutional model TouchPose learns to reconstruct a 3D hand skeleton, predict depth maps, and classify touch from low-resolution capacitive images; it trains on a cross-modal dataset of 65,374 aligned records collected with a mutual-capacitance touch panel, 16-bit digitizer, Leap Motion and time-of-flight depth camera. The system achieves 19.6 mm mean endpoint error (14.4 mm for contacting joints), generalizes to unseen poses and orientations, and enables intent detection and gesture-driven UI controls.
AI-written summary · about
related talks
07mFingerTrak: Continuous 3D Hand Pose Tracking by Deep Learning Hand Silhouettes Captured by ...
14mBack-Hand-Pose: 3D Hand Pose Estimation for a Wrist-worn Camera via Dorsum Deformation Network
10mProactive Sensing for Improving Hand Pose Estimation
14mDeepFisheye: Near-Surface Multi-Finger Tracking Technology Using Fisheye Camera
20mToward Realistic Hands Gesture Interface: Keeping it Simple for Developers and Machines
15mMeCap: Whole-Body Digitization for Low-Cost VR/AR Headsets
21mSketchingWithHands: 3D Sketching Handheld Products with First-Person Hand Posture
11mEfficient Finger Model and Accurate Tracking for Hover-and-Touch, Mid-air and Microgesture Intera...