A Webserver to detect ads on TV

Build with Neural Networks, Django and Vue

Note: Cloning this repository may take a while, because it's very large

It's only testet with python 3.6, other versions may not work properly All dependencys for python are in requirements.txt listed.

If you want to use the webserver, Redis must also be installed.

So that evrything works correctly the dateset and the saved network, must be downloaded. For the dateset go here and for the network go here

Go to the directory deepnet/examples for examples.

To start the server run sh deploy.sh.

If you prefer docker run docker-compose up

There are 2 parts to the app. The first part is the neural network and the algorithmen. The second part is the webserver.

The main part is the deepnet directory. It holds a self written deep neural network library.

The numpywrapper module is just so the neural network library can switch between numpy and cupy

The main parts are: app, src, vuedj

vuedj: The Django settings

app: Backend with Websocket

src: Frontend implemented with VueJS

thoughts is the directory with the Protocoll and the maturaarbeit

update_handler.py is the intersection between the server and the algorithmen