2025
On-device vision model on the ESP32-S3 Sense that detects food in wearable camera images. An on-device filter for passive nutrition logging.
Trained and deployed a lightweight TinyML image classification model on the ESP32-S3 Sense to detect whether food was present in wearable camera images.
Designed as an on-device filtering layer for passive nutrition logging, reducing the need to store or upload irrelevant images. The project pushed on embedded vision constraints: model size, latency, power consumption, and privacy-preserving wearable AI.