Vehicle Detection and Tracking using YOLOv3

This project demonstrates vehicle counting and classification using OpenCV and the YOLOv3(You Only Look Once) object detection model. It can analyze both real-time video streams to count and classify vehicles.

Table of Contents

• Requirements
• Usage
• Project Structure
• License

Requirements

• Python 3.x
• OpenCV
• NumPy
• YOLOv3 model configuration and weights files(YOLOv3-320 cfg and weights) (Download from YOLO website)
• 'coco.names' file containing the class names

Real-time Vehicle Counting and Classification

1. Place the video file in the video_file folder or update the video file path in the code.
2. Ensure you have the required model files and 'coco.names' file.
3. Run the realTime() function in the provided Python script.
4. Press 'q' to exit the real-time analysis.

Project Structure

• main: The main Python script for vehicle counting and classification.

• Utility:

  • tracker.py: The Euclidean Distance Tracker module for object tracking.
  • counting_vehicles.py: Function for counting vehicles
  • detected_objects.py: Function for finding the detected objects from the network output
  • find_center.py: finding center of bounding boxes

• model_config

  • yolov3-320.cfg and yolov3-320.weights: YOLOv3 model configuration and weights files.

• video_file

  • video.mp4: Sample video file for real-time analysis.

• coco_class_index

  • coco.names: File containing COCO dataset class names.

License

This project is licensed under the MIT License - see the LICENSE file for details.