Skip to content

Latest commit

 

History

History
97 lines (56 loc) · 2.41 KB

README.md

File metadata and controls

97 lines (56 loc) · 2.41 KB

Neural Style Transfert(NST)


This project is an implementation of Neural Style Transfer using PyTorch. Neural Style Transfer is a deep learning technique that merges the style of one image with the content of another, creating visually striking results by blending artistic features with photographic details.

What is NST ?

Technical specifications

Property Value
Framework PyTorch
Device MAC M2
Optimizer LBFGS
Time for 1 image ~20 min

How does it work ?

Extract characteristics with VGG19

The VGG19 pre-trained model is a convolutional neural network with 19 layers, including 16 convolutional layers and 3 fully connected layers, featuring 3x3 convolutional filters and max pooling, totaling approximately 143.67 million parameters for image classification tasks.

we start by extracting the characteristics of content and style

Optimization Loop

The ititial image can be:
  • content
  • style
  • random noise

Loss Function

$$ L_{\text{total}} = \alpha L_{\text{content}} + \beta L_{\text{style}} + \gamma L_{\text{TV}} $$

  • L_Content is the content loss.
  • L_Style is the style loss.
  • L_TV is the total variation loss.
  • alpha, beta, and gamma are hyperparamters for each loss.

Style Loss

Content Loss

Paper:Neural Style Transfer: A Review