UnderWatch
Edge AI
Computer Vision
Machine Learning
Arduino
Python
Elder monitoring system that uses local edge AI and Computer Vision built at IrvineHacks 2027. Uses an image categorization ML model that reached an accuracy of 83%. Uses confidence scoring for fall detection.
Overview
UnderWatch is a fall detection system built at IrvineHacks 2026, leveraging local edge AI and Computer Vision to enhance safety and autonomy.
Key features include:
- Custom-trained image categorization ML model with 83% accuracy.
- Real-time confidence scoring for reliable fall detection.
- Local edge processing for privacy and speed.
My contributions included programming the Arduino Uno Q to connect to the USB webcam, training the ML model using Edge Impulse, and integrating the system into a cohesive solution for elder monitoring.
Gallery
Key Specifications
Category
Edge AI
Role
Lead Developer