Classification service

Service for image classification with PyTorch, Flask, Redis and Gunicorn.

Classifier uses the Resnet18, trained on ImageNet (1000 classes).

Run

1. Install Docker.
2. Install Docker Compose.
3. Build and run services docker-compose up --build.

How to use

The service provides an asynchronous API.

1. Send POST request for classification
   curl "http://0.0.0.0:8000/classify" -F "file=@my_image.jpg"
   If the file is successfully read by the system, you will receive a response in which there is a uuid key.

2. Send GET request for checking result
   curl "http://0.0.0.0:8000/classify?uuid=<your_uuid>"
   Use uuid obtained on previous step, e.g. curl "http://0.0.0.0:8000/classify?uuid=b48b2506-962c-4592-a480-ed7756b88dde". You will receive the result of the prediction, or the status of the request processing.