🤖 Computer Vision & HCI Projects [Python]

A collection of advanced computer vision projects focused on real-time human interaction and safety monitoring systems.

Project Overview

This page showcases two different computer vision projects that demonstrate advanced real-time human interaction systems using OpenCV and MediaPipe.

1. Multimodal Gesture Recognition

A comprehensive gesture recognition system that processes multiple input modalities simultaneously.

Key Features:

• Hand gesture tracking and classification
• Facial expression analysis
• Body posture detection
• Multimodal gesture fusion
• Rhythmic movement detection

Supported Gestures:

• Hand Gestures: Peace sign, OK gesture
• Facial Expressions: Smile detection, eyebrow tracking
• Body Postures: Shrug gesture
• Complex Gestures: Thinking pose (multimodal)
• Rhythmic Gestures: Groove detection

2. Somnolence Detection System

A safety-focused application that monitors driver alertness through real-time eye tracking.

Key Features:

• Real-time eye tracking using MediaPipe Face Mesh
• Eye Aspect Ratio (EAR) calculation for drowsiness detection
• Visual drowsiness alerts with configurable thresholds
• Live monitoring dashboard
• Mirror display for user comfort

Technical Highlights:

• 6-point eye landmark detection
• Dynamic threshold-based alert system
• Configurable sensitivity settings
• Real-time performance optimization

Technical Requirements

Common Requirements:

• Python 3.9+
• OpenCV (cv2)
• MediaPipe
• NumPy
• SciPy
• Functional webcam

Optional:

• OpenCV-compatible GPU (improves performance)

Installation and Setup

Gesture Recognition:

git clone https://github.com/imprvhub/multimodal-gesture-recognition.git
cd multimodal-gesture-recognition
pip install -r requirements.txt
python gesture_recognition.py

Somnolence Detection:

git clone https://github.com/imprvhub/somnolence-detection.git
cd somnolence-detection
pip install -r requirements.txt
python somnolence_detection.py

Usage Instructions

Gesture Recognition System:

• Launches in fullscreen mode
• Controls:
  • ‘q’ - Quit
  • ’r’ - Reset calibration
  • ’esc’ - Exit fullscreen
• Displays real-time gesture recognition status

Somnolence Detection System:

• Application launches with webcam activation
• Press ‘q’ to quit
• Visual indicators show:
  • Green eye contours
  • EAR value display
  • Drowsiness warnings

Future Development

Gesture Recognition:

• Additional multimodal gestures
• Custom gesture training interface
• Gesture sequence detection
• Enhanced rhythm analysis

Somnolence Detection:

• Audio alert implementation
• Data logging capabilities
• Multiple face tracking
• Mobile optimization

Technical Architecture

Both projects feature:

• Multi-threaded processing
• Real-time optimization
• Modular design patterns
• Adaptive threshold systems
• State management
• Performance monitoring

Research Applications

These projects are designed for:

• Academic research
• Human-computer interaction studies
• Computer vision development
• Safety system prototyping
• Technical demonstrations

License

Both projects are released under the MIT License. See respective repositories for detailed terms.

---

Built with OpenCV and MediaPipe