👏 Clap Detector

A real-time clap detection system using pose estimation (YOLO) and wrist movement analysis, visualized live with PyQtGraph.

How It Works

The app uses a YOLO pose model to track wrist keypoints via webcam. It measures the normalized distance between wrists (relative to shoulder width) over a rolling window of frames, then computes signed and unsigned derivatives to detect clapping motion. A score above 0.7 triggers a clap detection.

Requirements

• Python 3.9+
• ultralytics
• torch
• pyqtgraph
• PySide6

Install dependencies:

pip install ultralytics torch pyqtgraph PySide6

Usage

python detect.py

Make sure yolo26n-pose.pt is in the same directory. A webcam feed will open alongside a live graph of wrist movement derivatives.

Configuration

Variable	Default	Description
SAMPLES	50	Rolling window size for wrist data
T	3	Frame window for derivative calc
TM	30	Graph history multiplier

Output

• Red line — signed dx (net wrist direction)
• Green line — unsigned dx (total wrist movement)
• Console prints CLAPPING: <score> when a clap is detected