EyeMouse: Gaze-Based Cursor Control System
CoFounder @ CHI Labs (IIT Kanpur) - 2011
Overview
Built a real-time gaze-driven cursor control system from scratch, without relying on neural nets or heavyweight CV libraries. The user simply looks at a point on screen, and the system maps head and eye landmarks via geometric models to move the cursor there with sub-100 ms latency on modest CPU hardware.
Design
- Custom Landmark Detection: Implemented low-latency iris and head-pose estimation using handcrafted geometric algorithms.
- Lightweight Pipeline: Optimized image capture, feature extraction, and mapping routines to run at ≥15 FPS on a single core, without GPU acceleration.
- End-to-End Integration: Connected video input, gaze estimation, and OS‐level cursor APIs into a turnkey desktop application.
Awards & Impact - IBM WebContest Innovation Award for “Best Student Project” (2011)
- Sparked my first hardware-software startup, partnering with an electronics engineer to build interactive assistive devices.
Resources