Telegram Bot @stylizer_bot providing image style transfer using NST & CycleGAN

Developed as a final project of the Deep Learning Course 1st part by MIPT Deep Learning School.

Currently deployed on Heroku Free Dyno so stylizing is quite slow and images in result are not large enough. Also waking the bot up by 1st message may take a couple of minutes.

Neural Style Transfer model is adopted from Alexis Jacq and rewritten by me.

CycleGAN model is a binding to the official implementation.

Try out the bot: @stylizer_bot

Contact me: @oleg_navolotsky