/Particle-filtering-for-visual-face-tracking-description-and-evaluation

Primary LanguagePython

Face Tracking with Particle Filtering

Overview

This Python script implements a particle filter-based hand gesture tracking algorithm using OpenCV. It tracks a face in a video stream by predicting and updating the state of particles representing potential face positions.

Features

  1. Particle Filter Implementation

    • Uses a particle filter for dynamic state estimation in a video stream.
    • Handles face tracking through particle representation.

  2. Real-Time Tracking

    • Performs real-time tracking of a face with bounding box visualization.

  3. User Interaction

    • Allows manual selection of the initial region of interest (ROI) around the face.

  4. Robust and Adaptable

    • Capable of adapting to occlusions and varying face appearances.

Usage

  1. Run the script.
  2. Select the face ROI in the first frame of the video. Screenshot
  3. Press 'Enter' to start tracking.
  4. Press 'q' to quit.

Limitations

  • Higher computational cost with increased number of particles.
  • Sensitive to initialization and parameter settings.
  • Primarily focused on face tracking, limiting its use in broader contexts.

Example Outputs

Screenshot