Ainize-run-wonny-example

This repository provides an API server, called Wonny server, that evaluates images based on an aesthetic and technical image quality model. The model used in the server is from Image Quality Assessment, which is an implementation of Google's research paper "NIMA: Neural Image Assessment".

The evaluation using Wonny server is done in the following steps:

1. User publishes an image file
2. User queries Wonny server with the image url
3. Wonny server returns a quality score of the image in the scale of 1 to 10

Note that the server is implemented in Node.js.

How to deploy

Wonny server is dockerized, so it can be built and run using docker commands.

Docker build

docker build -t wonny -f Dockerfile-cpu .

or

docker build -t wonny -f Dockerfile .

Docker run

docker run -p 80:80 -it wonny

Now the server is available at http://localhost. To learn how to query the server, see the next section.

Note that the docker image can be deployed using any docker-based deploy platform (e.g. ainize.ai).

How to query

How to publish an image file

The image to be evaluated needs to be published first. You can refer to the two following examples of how to publish image files:

Imgur

https://imgur.com/

Upload your image and get link.

Google Drive

For your convience, we share a Google Drive folder.

Once uploaded an image file, you can get a sharable link to the uploaded file from the image id provided by Google Drive:

https://drive.google.com/uc?export=view&id=${imageId}

How to query the server

Input Link for image.(Size of file should be less than 40Mb)

References

• Introducing NIMA: Neural Image Assessment on Google AI Blog
• https://github.com/idealo/image-quality-assessment