Advancing Hand Gesture Recognition with High Resolution Electrical Impedance Tomography
summary
A wireless, battery-powered armband prototype uses electrical impedance tomography with four-pole sensing and up to 32 electrodes to image the wrist; custom hardware raises sampling from ~250 to ~3000 measurements/sec and runs a single-step MAP reconstructor in real time. Evaluation uses saline phantoms and a 10-person user study (11 gestures, 3,300 trials). Four-pole outperforms two-pole (roughly 84% vs 94.3% peak gesture accuracy at 32 electrodes), showing higher electrode counts and four-pole drive materially improve recognition.
AI-written summary · about
chapters (19)
- 0:00Intro
- 0:20Tomography
- 0:46Tomography Techniques
- 2:00AIT
- 2:41Tomo
- 7:53New Hardware
- 8:07Schematic Diagram
- 8:40New Prototype
- 9:04Image Reconstruction
- 10:03Experiments
- 10:47Results
- 12:43RealTime Reconstruction
- 13:47Gesture Accuracy
- 14:07Data Collection
- 14:20Machine Learning
- 15:12Machine Learning Results
- 15:39Future Work
- 16:57EMG vs Sensing
- 18:22Other Questions
related talks
17mEMPress: Practical Hand Gesture Classification with Wrist-Mounted EMG and Pressure Sensing
19mBackHand: Sensing Hand Gestures via Back of the Hand
06mWR-Hand: Wearable Armband can Track User's Hand
14mLeveraging Dual-Observable Input for Fine-Grained Thumb Interaction Using Forearm EMG
05mA Feature Adaptive Learning Method for High-Density sEMG-Based Gesture Recognition
06mReal-time Arm Gesture Recognition in Smart Home Scenarios via Millimeter Wave Sensing
14mBack-Hand-Pose: 3D Hand Pose Estimation for a Wrist-worn Camera via Dorsum Deformation Network
28mStrike a Pose - Gesture recognition in JavaScript w/ ML & Arduino
