← All systems
P–07 / 07

Case study / Computer vision

Hand Gesture Puzzle Game

An experimental gesture-controlled interface connecting browser logic with camera input.

CategoryComputer visionStackPython · OpenCV · MediaPipeLicenseMITSourceGitHub ↗

Overview

What the system is

A webcam-based puzzle game controlled with real-time hand gestures: pinch to grab a shape, drag it across the screen, and drop it onto a matching moving target across five progressive levels.

Capabilities

What it does

  • 01Real-time hand trackingMediaPipe Hands reads webcam landmarks frame by frame.
  • 02Pinch to grabThumb–index distance, normalized by hand size, drives intuitive grab and drop.
  • 035 progressive levelsFrom static targets to smaller shapes with fast diagonal XY movement.
  • 04Live HUDHand-speed meter and FPS display with in-game restart and next-level controls.
  • 05Tunable gameplayPinch sensitivity, cursor smoothing, snap distance, and target speed configurable in config.py.

Engineering

Decisions on record

Vision
OpenCV capture with MediaPipe Hands (pinned 0.10.21 for the legacy solutions API).
Structure
Split across main, config, game, vision, and renderer modules.
Runtime
Python 3.11 with a webcam; cross-platform on Windows / macOS / Linux.

Credits: Srun Sochettra, Tep Makara & Sar Chanrithy