Skip to content

BrianSantoso/SketchGAN

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

SketchGAN

Deep Convolutional Generative Adversarial Network (DCGAN) for generating pencil sketches

Walking Through Latent Space:

(Linear interpolation between latent vectors)

The model contains a total of 24,216,196 trainable parameters.

The final layer of the generator network utilizes a sigmoid function that is stretched by a factor of 5 (by scaling the input by 0.2). This allows the model to more easily learn smooth pencil-like textures (as opposed to crisp, pen-like figures like the digits from the MNIST dataset).

The training set consists of 1001 unique grayscale images of pencil sketches of birds, horses, butterflies, disney characters, male and female faces, flowers, hearts, hands, still life, and other miscellanious objects. The image data was also flipped across the vertical axis to produce a total of 2002 pencil sketches.

Miscellanious sketches at random training iterations:

Addition of latent space vectors:

About

Deep Convolutional Generative Adversarial Network (DCGAN) for generating pencil sketches

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages