EchoFlex: Hand Gesture Recognition using Ultrasound Imaging
summary
Wearable forearm ultrasound images classify 10 discrete hand gestures using dense optical flow features fed to neural networks, achieving above 98% accuracy (10-fold CV on 12 participants). Continuous per‑finger flexion is regressed with <5% NRMSE using a bend‑sensor glove for ground truth. A simple optical‑flow realignment mitigates cross‑session probe shifts but the longitudinal probe plane degrades robustness. Practical limits remain: acoustic coupling (gel/pads), device size, power and cost.
AI-written summary · about
chapters (10)
- 0:00Introduction to study
- 0:35Background and context
- 2:35Ultrasound imaging basics
- 6:20User study and methodology
- 8:09Sensor misalignment solution
- 9:29Continuous finger tracking
- 10:41Advantages of ultrasound
- 11:21Practical implementation issues
- 12:33Future directions and summary
- 14:46Q&A session
related talks
17mEMPress: Practical Hand Gesture Classification with Wrist-Mounted EMG and Pressure Sensing
06mDesigning and Evaluating Hand-to-Hand Gestures with Dual Commodity Wrist-Worn Devices
20mTomo: Wearable, Low-Cost Electrical Impedance Tomography for Hand Gesture Recognition
07mFingerTrak: Continuous 3D Hand Pose Tracking by Deep Learning Hand Silhouettes Captured by ...
18mCyclopsRing: Enabling Whole-Hand and Context-Aware Interactions Through a Fisheye Ring
05mReady, Steady, Touch! — Sensing Physical Contact with a Finger-Mounted IMU
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