Environment

• Ubuntu 18.04 or higher
• Python 3.7.5

Requirements

• ffmpeg
• PyTorch 1.7+
• pip install -r requirements.txt

Download YOLOv3 pretrained weight

mkdir yolo_weights
wget -P yolo_weights https://pjreddie.com/media/files/yolov3.weights
wget -P yolo_weights https://raw.githubusercontent.com/pjreddie/darknet/master/cfg/yolov3.cfg
wget -P yolo_weights https://raw.githubusercontent.com/pjreddie/darknet/master/data/coco.names
wget -P yolo_weights https://raw.githubusercontent.com/adipandas/multi-object-tracker/master/examples/pretrained_models/yolo_weights/coco_names.json

Steps

1. Go to project directory
2. Execute python http_server
3. Execute python server.py
4. Execute python stream.py
5. Open http://<server_ip>:<http_port>/index.html (<server_ip> and http_port are defined in config.json).

Reference

• UI design:
  • https://freshman.tech/custom-html5-video/
  • https://www.w3schools.com/css/css_dropdowns.asp
• HLS streaming: https://video.aminyazdanpanah.com/python/start
• Flask server: https://towardsdatascience.com/video-streaming-in-web-browsers-with-opencv-flask-93a38846fe00