Skip to content

facial_expression_analyzer_README.md

Latest commit

 

History

History
106 lines (69 loc) · 3.5 KB

facial_expression_analyzer_README.md

File metadata and controls

106 lines (69 loc) · 3.5 KB
Raw

Facial Expression Analyzer

This project contains scripts for real-time facial expression analysis using a webcam. The scripts detect faces and analyze facial expressions, displaying the results in a live video feed.

Scripts

1. facial_expression_analyzer.py

The original implementation that uses the FER (Facial Expression Recognition) library. This script may require additional dependencies and permissions.

Note: This script has dependency issues with the moviepy.editor module and may not work without additional configuration.

2. facial_expression_analyzer_simple.py

A simplified version that uses OpenCV's Haar Cascade classifier for face detection. This script only detects faces without analyzing expressions.

Features:

  • Face detection using OpenCV's Haar Cascade classifier
  • Real-time video feed with face rectangles
  • FPS counter
  • Simple user interface

3. facial_expression_analyzer_advanced.py (Recommended)

An advanced implementation that uses OpenCV for face detection and a simple heuristic-based approach for emotion classification.

Features:

  • Face detection using OpenCV's Haar Cascade classifier
  • Basic emotion classification (Happy, Sad, Angry, Surprise, Neutral)
  • Color-coded face rectangles based on detected emotion
  • Real-time video feed with emotion labels
  • FPS counter
  • Screenshot capability (press 's' to save)
  • Clear user interface with instructions

Requirements

  • Python 3.6+
  • OpenCV (opencv-python)
  • NumPy
  • Matplotlib (for visualization)

Installation

  1. Ensure you have Python installed
  2. Install the required packages:
pip install opencv-python numpy matplotlib

Usage

Running the Advanced Analyzer

python facial_expression_analyzer_advanced.py

Command-line Options

--camera INT       Camera device ID (default: 0)
--confidence FLOAT Confidence threshold for face detection (0.0-1.0)
--debug            Enable debug logging

Controls

  • Press 'q' to quit the application
  • Press 's' to save a screenshot (advanced version only)

Webcam Permissions

These scripts require webcam access. On macOS, you may need to grant permission in: System Preferences > Security & Privacy > Privacy > Camera

How It Works

Face Detection

The scripts use OpenCV's Haar Cascade classifier to detect faces in the video feed. This is a machine learning-based approach that uses a cascade of simple features to identify faces.

Emotion Classification (Advanced Version)

The advanced version uses a simple heuristic-based approach to classify emotions based on pixel intensity in different regions of the face (eyes, mouth, forehead). This is a simplified demonstration and not a production-ready emotion classifier.

In a real-world application, you would use a trained deep learning model for more accurate emotion classification.

Screenshots

Screenshots are saved in the current directory with filenames like facial_expression_20250413_093849.jpg (timestamp format).

Limitations

  • The emotion classification in the advanced version is based on simple heuristics and is not highly accurate
  • Face detection may struggle in poor lighting conditions or with unusual face angles
  • The scripts require a webcam with proper permissions

Future Improvements

  • Implement a proper deep learning-based emotion classifier
  • Add facial landmark detection for more detailed analysis
  • Improve the UI with more detailed emotion metrics
  • Add recording capability for saving video
  • Implement eye blink detection and attention monitoring