🚗 YOLO11 Vehicle Detection from YouTube Stream

This Python project uses YOLO11 for real-time vehicle detection from a YouTube video stream. It tracks vehicles like cars and displays the total number of vehicles passed along with the elapsed time on the video stream.

📷 Demo

✨ Features

• Real-time vehicle detection using YOLO11.
• Detects cars from YouTube live streams and videos.
• Tracks detected vehicles using SORT (Simple Online and Realtime Tracking).
• Displays bounding boxes and vehicle counts on the video stream.
• Shows elapsed time and vehicle counts on the video.

🛠️ Tech Stack

• Python
• YOLO11
• OpenCV
• yt-dlp for extracting YouTube video streams
• SORT for tracking objects
• NumPy

🚀 Quick Start

1. Clone the repository:

   git clone https://github.com/pablocaeg/opencv-traffic-detector.git
   cd opencv-traffic-detector

2. Install dependencies:

   pip install filterpy scikit-image numpy opencv-python torch ultralytics yt-dlp

3. Install SORT:

   SORT isn't available via pip, so you need to install it manually:

   git clone https://github.com/abewley/sort.git
   cd sort
   pip install -r requirements.txt
   python setup.py install

4. Run the script:

   python3 detection.py

5. Replace the YouTube URL:

   • In detection.py, replace the youtube_url variable with the YouTube video or livestream URL you want to process:

   youtube_url = 'https://www.youtube.com/watch?v=YOUR_VIDEO_ID'

📊 Stats Displayed on Video

• Total Cars Passed: Counts and displays the total number of cars detected during the video stream.
• Elapsed Time: Shows how long the script has been running in HH:MM:SS format.

🖼️ Example Output

When the script is running, you'll see a video stream with bounding boxes around the detected cars and real-time stats overlaying the video.

Example of stats:

Cars Passed: 14
Time Elapsed: 00:12:43