merve HF staff commited on
Commit
3afffa1
1 Parent(s): 83f6c16

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +17 -0
README.md CHANGED
@@ -3,3 +3,20 @@ library_name: keras
3
  tags:
4
  - image-to-image
5
  ---
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
3
  tags:
4
  - image-to-image
5
  ---
6
+ # Conditional Generative Adversarial Network
7
+ This repo contains the model and the notebook to [this Keras example on Conditional GAN](https://keras.io/examples/generative/conditional_gan/).
8
+
9
+ Full credits to: [Sayak Paul](https://twitter.com/RisingSayak)
10
+
11
+ # Background Information
12
+
13
+ Training a GAN conditioned on class labels to generate handwritten digits.
14
+
15
+ Generative Adversarial Networks (GANs) let us generate novel image data, video data, or audio data from a random input. Typically, the random input is sampled from a normal distribution, before going through a series of transformations that turn it into something plausible (image, video, audio, etc.).
16
+
17
+ However, a simple DCGAN doesn't let us control the appearance (e.g. class) of the samples we're generating. For instance, with a GAN that generates MNIST handwritten digits, a simple DCGAN wouldn't let us choose the class of digits we're generating. To be able to control what we generate, we need to condition the GAN output on a semantic input, such as the class of an image.
18
+
19
+ In this example, we'll build a Conditional GAN that can generate MNIST handwritten digits conditioned on a given class. Such a model can have various useful applications:
20
+
21
+ let's say you are dealing with an imbalanced image dataset, and you'd like to gather more examples for the skewed class to balance the dataset. Data collection can be a costly process on its own. You could instead train a Conditional GAN and use it to generate novel images for the class that needs balancing.
22
+ Since the generator learns to associate the generated samples with the class labels, its representations can also be used for other downstream tasks.