A real-time gesture-controlled application built at SpurHacks 2025. Uses OpenCV and MediaPipe for hand tracking, Gemini AI for intelligent responses, all served through a Flask backend.
Built in 36 hours at SpurHacks 2025, this application enables real-time gesture-based interaction with an AI system. The computer vision pipeline uses MediaPipe's hand landmark detection to track 21 hand keypoints at 30fps, processed through OpenCV for gesture recognition.
Recognized gestures are mapped to commands and sent to Google's Gemini AI API for intelligent contextual responses. The Flask backend orchestrates the entire pipeline, handling video stream processing, gesture classification, and API communication.
Features include custom gesture training, multi-hand support, and a responsive web interface for viewing the camera feed with landmark overlay.