This script uses the YOLOv3 object detection model to detect objects in a video stream. It can be used with any video stream, as long as the URL of the stream is specified in the script.
To run this script, you will need to have the following dependencies installed:
- OpenCV
- NumPy
You can install these dependencies using pip:
pip install opencv-python numpy
You will also need to download the YOLOv3 configuration file and weights from the official YOLO website:
Make sure that the configuration file and weights are located in the same directory as your Python script.
Before running the script, you will need to start a video stream using ffmpeg. You can use the following command to start a video stream from a webcam and stream it to a UDP address:
ffmpeg -f v4l2 -i /dev/video0 -preset ultrafast -tune zerolatency -f mpegts udp://192.168.5.185:1234
This command starts a video stream from /dev/video0 (your webcam) and streams it to udp://192.168.5.185:1234.
To use this script, you will need to modify the url variable in the script to match the URL of the video stream you want to use. In this case, the URL should be:
url = 'udp://192.168.5.185:1234'
You can also modify the conf_threshold and iou_threshold variables to adjust the detection sensitivity and non-maximum suppression overlap threshold, respectively.
Once you have modified the variables, you can run the script using the following command:
python main.py
This will open the video stream and start detecting objects in real-time. Detected objects will be highlighted with bounding boxes and labeled with their class and confidence score. To exit the script, press the 'q' key.
This script uses the pre-trained YOLOv3 model, which was developed by the Darknet team. The original YOLOv3 paper can be found here.