Gauntlet: Gesture-Controlled Workspace Launcher
A hands-free Windows workspace launcher that recognizes a snap gesture and arranges four productivity applications across two displays.
- Python
- OpenCV
- MediaPipe
- Computer vision
Overview
Gauntlet is a gesture-controlled workspace launcher built with Python, OpenCV, and MediaPipe. It recognizes a snap gesture and arranges four productivity applications across two displays.
My Role
I developed a hands-free Windows workspace launcher and implemented real-time webcam processing with OpenCV and MediaPipe.
Inspiration
Heading into university, I was looking for ways to optimize my workflow and simplify setting up my workspace. The name takes inspiration from Thanos' gauntlet bringing multiple powers together in one place. Gauntlet brings my productivity tools together through a single snap gesture.
Development Process
Built a real-time computer-vision pipeline using OpenCV and MediaPipe to validate webcam frames and extract 21 landmarks from each detected hand. I normalized the thumb-middle finger distance against hand length, then used a timed state machine with separate closing and opening thresholds to recognize deliberate snap gestures and reduce false detections from hand-tracking jitters.
Results
Automated the launch and placement of four Windows applications across two displays using pywin32 and psutil, triggered by a deliberate snap gesture.
What I Learned
The project gave me practical experience with Python, processing webcam frames, and extracting hand landmarks using OpenCV and MediaPipe.
Technologies
- Python
- OpenCV
- MediaPipe