Python YOLO Object Detection using OpenCV - Flask App

To run this locally:

• Download the model weights and put them in the app/weights. First create a weights directory in the app directory, then place the downloaded weights into this folder on your machine - these are the model weights, the app will not run without them. If you've successfully done this then you would have the following directory and file present app/weights/yolov3.weights.

• In the terminal:

  1. run the package dependencies pip install -r requirements.txt
  2. We want to run the flask app in development mode so run the following export FLASK_ENV=development
  3. export FLASK_APP=run.py
  4. To run the flask app flask run

• Go to http://127.0.0.1:5000/uploadfile to use the flask app

• Run Ctrl+ C in the terminal to kill the flask app

• To run this via Docker:

  1. Build the Docker image specified in the Dockerfile: docker build -t <name of image> . where <name of image> can be whatever you name the image - I suggest flask_yolo as an image name.
  2. To run the docker image (that doesn't persist): docker run --rm -p 5000:5000 <name of image>

• My Docker commands cheatsheet

• Warning: The Docker image may crash due to memory issues - to avoid this you can increase the docker memory. The instructions for Mac are here