vinibarbieri/vehicle-counting

Vehicle Counting with YOLO

This project uses the YOLO (You Only Look Once) object detection model to count vehicles in a video. The code processes each frame of the video to detect and track vehicles, providing a count of the detected objects within a defined region of interest (ROI).

Table of Contents

1. Features
2. Requirements
3. Installation
4. Usage
5. How It Works
6. Customization
7. Results

Features

• Detects and counts vehicles in a video using the YOLO model.
• Customizable region of interest (ROI) for counting objects.
• Visualizes object detection and tracking in real-time.
• Supports multiple object classes (e.g., vehicles, pedestrians).
• Outputs a processed video with counted objects and tracked trajectories.

Requirements

• Python 3.7+
• OpenCV
• Ultralytics YOLO

Installation

1. Clone this repository:

   git clone https://github.com/your-username/vehicle-counting-yolo.git

2. Navigate to the project directory:

   cd vehicle-counting-yolo

3. Install the required packages:

   pip install -r requirements.txt

Usage

1. Place your video file in the data/videos directory.

2. Modify the video_path variable in the code to point to your video file:

   cap = cv2.VideoCapture("data/videos/video.mp4")

3. Run the script:

   python vehicle_counting.py

4. The processed video will be saved as object_counting_output.avi in the project directory.

How It Works

1. Loading the Model: The pre-trained YOLO model is loaded to detect objects in the video frames.
2. Video Processing: The video is processed frame by frame to detect and track objects.
3. Object Counting: Objects that pass through the defined region of interest (ROI) are counted.
4. Output: The processed video with the counted objects and their tracked paths is saved and displayed.

Customization

• Region of Interest (ROI): Modify the line_points variable to change the area where objects are counted.

  line_points = [(170, 400), (1100, 400)]  # Define your ROI points here

• Object Classes: You can specify which object classes to count using the classes_to_count variable.

  classes_to_count = [0, 2]  # Modify the class IDs as needed

Results

The processed video will display:

• Bounding boxes around detected objects.
• Trajectories of tracked objects.
• Count of objects crossing the defined ROI.