Skip to content

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

13 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Camvas

Camvas turns your webcam feed into a hand-drawable canvas. Draw directly on your video feed with your hand and let the AI correct your drawing into a clean geometric shape.

The annotated feed is exposed as a virtual camera device, so any app (Zoom, Meet, Teams, Discord, Slack huddles, browser tabs) can use it as a webcam source.

The virtual camera is currently supported only on Windows (using Softcam )

How it works

Every webcam frame flows through a small pipeline of layers:

  1. Webcam Layer — Pulls the feed directly from your webcam.
  2. Hand tracking — a MediaPipe hand-landmark model locates your index-tip and thumb-tip in the frame.
  3. Gesture interpretation — if the two tips are close enough, that counts as a pinch. The pinch either hits an on-screen button (if the finger is over one) or starts / continues a drawing stroke.
  4. Classification — when the pinch releases and the stroke ends, the stroke canvas is cropped, binarized, resized to 28×28, and fed through a small CNN. The model returns a shape label and a confidence score.
  5. Correction — if the model is confident enough, the raw stroke is replaced with a clean geometric version (works by analysing vertices and edges). Uncertain strokes are kept as raw ink instead.
  6. Compositing — the output frame is built in layers: webcam image at the base, then finalized drawings, then any in-progress stroke, then the UI strip (buttons and label) at the top.
  7. Virtual-camera output — the composed frame is pushed to a virtual webcam device, which your video or streaming app reads as if it were a real camera.

Download

Prebuilt installer (Windows):

Download the latest release

Features

  • Works anywhere - Just switch the virtual camera in any app
  • Auto-shape-correction for squares, circles, triangles, lines, and arrows.
  • Toolbar - Displays a toolbar directly over your video feed with buttons that you literally "hold" to activate. Allows moving drawings, changing color, undo, and clear. Also supports shortcuts
  • Real-time performance - Optimised to run smoothly in real-time at high frame rate and minimal CPU usage.
  • Runs in background - Can run on startup and stay ready in background. Automatically clears memory when not in use.

Running locally

Prerequisites

git clone https://github.com/Divyesh06/Camvas.git
cd Camvas
uv venv
uv pip install -r requirements.txt
uv run poe dev

About

No description, website, or topics provided.

Resources

Stars

0 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages