SensIR: Detecting Hand Gestures with a Wearable Bracelet using Infrared Tomography
summary
A wearable bracelet is built with 14 LEDs and photodiodes (peak 860 nm) that multiplex every emitter against every receiver to form a brightness matrix fed to a neural network classifier. It detects 12 discrete gestures with 93.3% accuracy, uses a one-to-many emitter-receiver matrix instead of single pairs, and the prototype is a 14-segment bracelet that tolerates gaps, some rotation, and session misalignment with a simple calibration regressor.
AI-written summary · about