Image orientation detection

App inspired by the work of Deepanshu Thakur and Simon Willison

This webapp was created on behalf of the British Library Endangered Archive Programme to determine the orientation of images in their collection. The deep learning model is based on a 50-layer version of resnet and was trained on around 650 unique images from the EAP archives using the fast.ai library. Each image is duplicated into four representations, consisting of the application of 0, 90, 180 or 270 degrees of anti-clockwise rotation.

The 99MB pre-trained model is included in this package and is found within resnet50_EAP_version6.pkl

imgapp.py is a Starlette API server which accepts image uploads or image URLs and runs them against the pre-calculated model.

A Dockerfile is included for convenience.

Deployment

If deploying on a server, run the Starlette app on the uvicorn ASGI server behind an nginx reverse proxy

cd $APP_DIR
sudo apt-get install nginx
(sudo) cp nginx_config_file.conf /etc/nginx/sites-available/default
sudo ln -s /etc/nginx/sites-available/default /etc/nginx/sites-enabled/

Before this step, edit the nginx config file ip addresses to match your setup

Check the health of the config file settings:

sudo nginx -t

Reload nginx if all is ok:

sudo /etc/nginx/init.d/nginx reload

Automating things

Check the status of nginx

sudo systemctl status nginx

Start the server

sudo systemctl start nginx

Configure nginx to automatically start on system startup sudo systemctl enable nginx