• Wingman: Hazard Detection and Warning System for Cyclists

A low-cost computer vision-based real-time hazard detection system for cyclists, empowering riders with enhanced situational awareness at 10% of the cost of leading LIDAR solutions. Uses Kivy (mobile app GUI), the Arduino ESP32-CAM microcontroller (for rearward-facing video acquisition sent over wifi to the mobile device), a Python ML backend for object detection and classification using TensorFlow Lite and the MobileNet vision model plus a custom monocular distance estimation model trained using the KITTI Vision Dataset via XGBoost using an implementation from the DisNet 2018 Paper.

Wingman Product Mockup and Poster Wingman Bike, Arduino, and Phone Render Wingman Arduino Render

  • Real-Time Profanity Filtering in Audio

Low-latency profanity filter using the Google Cloud Speech-to-Text API with PyAudio and Python multiprocessing to manage input and output audio streams.

Profanity Icon