d3vv3 / autofocus-server

Server for autofocus

autofocus-server

Server for autofocus.

What is autofocus?

Autofocus server is a fast, light, scalable photo indexer with automatic face and object recognition.

How are we indexing?

We are indexing images by:

• Faces: using some fast 'face_recognition' library.
• Places: on a map and quering for names of location coordinates.
• Object recognition: using some YOLO v3 model or others.
• Others: tags will be added automatically or by hand (optional).

This implies if you search for red car you will get the results from the most
matches to the least (till no matches).

Performance

We are building a blazing fast and light server, in order to be able to run it
on as many computers as possible. Even if they are old.

Tools

We have chosen to build autofocus using:

• Python: the language I mostly speak.
• ReactJS: so our web is responsive, functional and scalable.
• Flutter: so our app feels native everywhere.

Obviously the tools are not everything. Code must be built efficiently.

Install (development)

1. Install face_recognition from here (dlib included).
2. Install requirements by pip install -r requirements.txt.
3. Deploy mongodb with docker run -d --name mongo -p 27017:27017 -e MONGO_INITDB_ROOT_USERNAME=autofocus -e MONGO_INITDB_ROOT_PASSWORD=apassword mongo
4. Edit the .env file for your setup.
5. Start developing!