REAL-TIME HAND TRACKING

REAL-TIME HAND TRACKING

Creative project | Computer vision | 2026

A real-time hand tracking desktop app for macOS built with OpenCV and MediaPipe. It captures the built-in webcam, detects up to two hands and overlays their landmarks and skeleton connections live in a mirrored, resizable window with an FPS counter and an on-screen quit button.

Key contributions:

Implemented live hand landmark detection and skeleton overlay with MediaPipe Hands, using distinct colours for the first and second detected hand and a bounding box around each.

Handled macOS camera selection through AVFoundation so the built-in camera is chosen over a connected iPhone (Continuity Camera).

Added a mirrored view, FPS counter and custom UI styling (SF Compact font, rounded button, configurable palette) rendered with Pillow.

Pinned the MediaPipe version deliberately after newer releases dropped the legacy solutions API, and documented the reasoning for future maintenance.

Collaborators: Gabriela Słomiany

Software Used: PyCharm, OpenCV, MediaPipe, Pillow

Programming Languages: Python

Repository: github.com/imgabrielas/cv-hand-tracking

Project Date: July 2026