Real-Time Neural Style Transfer : Using Deep Learning to Generate Art

In this project we delve into the realm of Neural Style Transfer (NST) which’s a technique that uses deep learning to infuse an image with the artistic style of another. Typically NST has been computationally intensive which has restricted its capabilities. However this project seeks to overcome this challenge by harnessing the potential of MSG Net (Multi Style Generative Network) a learning framework specifically developed for real time NST. Using the capabilities of MSG Net we can unleash the power of Neural Style Transfer (NST) in real time.

Just picture being able to upload a portrait and instantly witnessing it undergo a transformation, into a masterpiece inspired by Van Goghs style.

TO RUN THE CODE:

1. Extract the model: cd experimens/main.py

2. To run the camera_demo, specify the model and disable cuda: Run camera_demo.py

3. Run python camera_demo.py demo --model models/21styles.model --cuda=0