Yolov5 object detection model deployment using flask

This repo contains example apps for exposing the yolo5 object detection model from pytorch hub via a flask api/app.

Web app

Simple app consisting of a form where you can upload an image, and see the inference result of the model in the browser. Run:

$ python3 webapp.py --port 5000

then visit http://localhost:5000/ in your browser:

Processed images are saved in the static directory with a datetime for the filename.

Rest API

Simple rest API exposing the model for consumption by another service. Run:

$ python3 restapi.py --port 5000 --model yolov5s

Then use curl to perform a request:

$ curl -X POST -F image=@tests/zidane.jpg 'http://localhost:5000/v1/object-detection/yolov5'

The model inference results are returned:

[{'class': 0,
  'confidence': 0.8197850585,
  'name': 'person',
  'xmax': 1159.1403808594,
  'xmin': 750.912902832,
  'ymax': 711.2583007812,
  'ymin': 44.0350036621},
 {'class': 0,
  'confidence': 0.5667674541,
  'name': 'person',
  'xmax': 1065.5523681641,
  'xmin': 116.0448303223,
  'ymax': 713.8904418945,
  'ymin': 198.4603881836},
 {'class': 27,
  'confidence': 0.5661227107,
  'name': 'tie',
  'xmax': 516.7975463867,
  'xmin': 416.6880187988,
  'ymax': 717.0524902344,
  'ymin': 429.2020568848}]

Run & Develop locally

Run locally for dev, requirements mostly originate from yolov5:

python3 -m venv venv
source venv/bin/activate
(venv) $ pip install -r requirements.txt
(venv) $ python3 restapi.py --port 5000

An example python script to perform inference using requests is given in tests/test_request.py

Docker

The example dockerfile shows how to expose the rest API:

# Build
docker build -t yolov5-flask .
# Run
docker run -p 5000:5000 yolov5-flask:latest

reference