Style Transfer using CNN and VGG16 Architecture

Course project for CSE445 Introduction to Machine Learning

Summer 2017, North South University

Dependencies

• python3
• scipy
• keras
• tensorflow / tensorflow-gpu
• pillow
• numpy

Usage

• You can run using the accompanying IPython notebooks in Work/notebook folder
• Or you can run Work/py_scripts/style_transfer.py like

python style_transfer.py path/to/style/image path/to/content/image

Note that outputs will be faster on a machine with CUDA supported Nvidia GPU

• From naming - style image supplies the style to be applied on a content or a target image

Demo outputs

References

References used in this project can be found here