The Problem
Traditional digital painting tools depend on physical input peripherals such as mice, trackpads, or graphic tablets. Creating an accessible, touchless drawing experience using standard webcams introduces key technical hurdles: tracking latency, coordinate jitter caused by natural hand tremors, false stroke triggers during repositioning, and seamless webcam streaming in modern browsers.
Architecture & Approach
The pipeline uses Google MediaPipe Hands to capture 21 3D hand landmarks in real time from webcam video. Geometric heuristics classify gestures into four active states: single index finger for drawing, open palm for circular erasing, thumb-index pinch for canvas panning, and closed fist for idle movement. Coordinate streams are filtered via exponential smoothing for jitter-free strokes. The Streamlit web UI provides live in-browser webcam streaming, neon palette presets, adjustable brush/eraser sizes, AR overlay mode, undo history, and PNG artwork export.
About the Project
An interactive, touchless hand-gesture drawing and painting system built with Streamlit, OpenCV, MediaPipe, and NumPy. Features real-time 21-landmark 3D hand tracking to draw, erase, and pan across a digital canvas with natural hand postures. Includes exponential pointer smoothing, interactive neon color palettes, AR overlay view, dynamic stroke sizing, undo history, and PNG artwork export.
Key Metrics
21 3D Landmarks
Hand Tracking
Exponential Moving Avg
Smoothing Algorithm
Draw · Erase · Pan · Idle
Gesture Modes
Streamlit Web + Desktop OpenCV
Dual Runtime