Smart devices running an ML model can be made to “see” and “wake up” to perform an action upon object or presence detection.
Why ML: Separates overlapping objects and tracks movement more accurately than traditional motion detection.
Machine Learning Low-Resolution Vision Resources
People Flow Counting
Counts and tracks people using low-resolution thermal vision and on-device ML. Enables privacy-preserving occupancy and people flow analytics.
Gesture Recognition + Face Tracking
Uses separate vision models for face detection and hand gesture recognition to enable touchless smart device interaction.
Rock Paper Scissor Classification
Recognizes rock, paper, and scissors hand gestures locally using embedded vision AI. Serves as a starting point for custom image classification applications.