yolo-api-service

This repository consists of an example of how to create a API service for your YOLO model (https://pjreddie.com/darknet/yolo/). This is a very simple API implementation where you upload an image and then a the detected objects will appear in text form. You can very easily modify this code to work as an API to suit your needs.

Prerequisites

1. Linux Environment
2. Python 3
3. Flask

Set-up

This code will only work in a Linux environment

1. Set-up YOLO Darknet from either https://pjreddie.com/darknet/yolo/ or https://github.com/AlexeyAB/darknet in a Linux based environment. Make sure that you either have the your trained model ready (both a .cfg file and a .weights file).
2. Copy the app.py file into the darknet repository from step 1
3. In the app.py file. Change the function default parameters in runModel to match your cfg file location and weights file location. For Example:

runModel(filename,cfgFile="cfg/yolov4.cfg", weightsFile="yolov4.weights")

Run the server locally

1. You can run python3 app.py to get your local debugging server ready.
2. Open up the main webpage at <your_ip_address:5000/predict-image>
3. You can now upload a image file and get the predicted objects detected.