CCTV Distance Estimation

This project uses a YOLOv8 segmentation model to estimate the distance of objects (persons and monitors) from a CCTV camera in real-time.

Features

• Object detection and segmentation: Detects and segments persons and monitors in CCTV footage using YOLOv8.
• Distance estimation: Calculates the distance of detected objects from the camera using the perceived width and known width of the objects.
• Real-time processing: Processes the CCTV stream in real-time to provide continuous distance estimations.
• Visualization: Displays the video feed with bounding boxes around detected objects and their estimated distances.

Requirements

• Python 3.8 or higher
• ultralytics
• opencv-python
• numpy

Installation

1. Clone the repository:

   git clone https://github.com/omukesh/yolov8_distance_calculator

2. Install dependencies:

   pip install -r requirements.txt

Configuration

1. CCTV URL:
   • Create a file named config.py in the project directory.
   • Add your CCTV URL to config.py:

     CCTV_URL = "your_cctv_url" 

2. Focal Length Calibration:
   • Run the calculate_focal_length method in distance_calculator.py.
   • Place a known-sized object at a known distance from the camera.
   • Measure the pixel width of the object in the captured frame.
   • Update the calculate_focal_length method with the measured pixel width and known values.

Usage

1. Run the application:

   python yolo_distance.py

2. The application will open a window displaying the CCTV footage with bounding boxes and distance estimations.
3. Press 'q' to exit the application.

Notes

• Ensure the class IDs for "Person" and "Monitor" are correct for your YOLOv8 model.
• Adjust the known_width value in distance_calculator.py if necessary.
• Accurate focal length calibration is crucial for accurate distance estimation.
• Performance may vary depending on hardware and scene complexity.

Contributing

Contributions are welcome! Feel free to open issues or submit pull requests.

License

This project is done for educational and research purpose.