Neural-Style-Tranfer-From-Scratch

I build a NST with VGG-19 from scratch

There are a lot of existing NST libraries/modules avaible for Python, but for the sake of understanding it and understanding how Deep CNN works in terms of identifing features, I build this project almost from scratch(here I downloaded pretrained data for my own model to increase efficiency and save training time)

Link to the pretrained VGG-19 model can be found here

For details/how it is built, please go to the notebook

Disclaim

I wrote all the code in the Notebook, but codes in nst_utils(which are only several helper methods) are borrowed online to increase efficiency

What does this project do?

and

equals

Can't open?

Cannot open notebook/To view on phone, click here