Skip to content

Neural Style Transfer using Tensorflow framework and VGG-19 model

Notifications You must be signed in to change notification settings

darelover/Neural-Style-Transfer

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

6 Commits
 
 
 
 
 
 

Repository files navigation

Neural-Style-Transfer

This is a Tensorflow implementation of styling images using VGG-19 model.

Some Examples

Content Image Style Image Generated Image
Macau Starry Night MacauStarryNight
Star Wars Feneck StarWarsFeneck
Star Wars Starry Night StarWarsStarryNight

Hyperparamter Tuning

Style Strength

Alpha Beta Result
0.025 5 Generated_11
1 500 Generated_12
5 100 Generated_13
10 40 Generated_14
7.5 100 Generated_15
10 5000 Generated_16

Content Formula

Here,

Jcontent is the content loss function
C and G represent content and generated image respectively
a(C)and a(G) are the activations from the chosen layer when image is passed to the model
nH, nW, nC are the height, weight and number of channels of the image respectively

Formula Used Generated Image(Run only upto 100 iterations)
First GeneratedFirst
Second GeneratedSecond
Third GeneratedThird

References:

  1. A Neural Algorithm of Artistic Style
  2. Convolutional Neural Networks by deeplearning.ai on coursera
  3. NeuralArt Github Repository

About

Neural Style Transfer using Tensorflow framework and VGG-19 model

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published