CODEWITHSUNDEEP
data-augmentationdcgan
Gabi Kubi
Gabi Kubi

Reputation: 1

How to Improve Image Quality in My DCGAN for Data Augmentation

I am currently training a DCGAN (Deep Convolutional Generative Adversarial Network) for data augmentation, and I've based my implementation on the network proposed in this GitHub repository: https://github.com/eriklindernoren/PyTorch-GAN/blob/master/implementations/dcgan/dcgan.py

However, I'm facing issues with the generated images not being well-defined, and I'm not sure how to adjust the network's parameters to improve the quality. Here are the parameters I'm currently using:

parser.add_argument("--n_epochs", type=int, default=200, help="number of epochs of training") parser.add_argument("--batch_size", type=int, default=64, help="size of the batches") parser.add_argument("--lr", type=float, default=0.0002, help="adam: learning rate") parser.add_argument("--b1", type=float, default=0.5, help="adam: decay of first order momentum of gradient") 
parser.add_argument("--b2", type=float, default=0.999, help="adam: decay of first order momentum of gradient")
parser.add_argument("--n_cpu", type=int, default=8, help="number of cpu threads to use during batch generation") 
parser.add_argument("--latent_dim", type=int, default=100, help="dimensionality of the latent space") parser.add_argument("--img_size", type=int, default=32, help="size of each image dimension") parser.add_argument("--channels", type=int, default=1, help="number of image channels") parser.add_argument("--sample_interval", type=int, default=400, help="interval between image sampling")

Issues Observed:

Generated images are not well-defined and lack clarity.
Adjusting the number of epochs and batch size has not significantly improved the image quality.

Questions:

What parameter adjustments can I make to improve the quality of the generated images?
Are there any specific techniques or strategies to enhance the definition of the images?
How should I monitor and interpret the loss functions for both the generator and the discriminator to ensure proper training?

Any help or suggestions would be greatly appreciated!

Thank you!

  1. Number of Epochs and Batch Size: I have experimented with increasing the number of epochs and adjusting the batch size, but this has not significantly improved the image quality.
  2. Learning Rate and Adam Parameters: I have tweaked the learning rate and the parameters b1 and b2 of the Adam optimizer, but I haven't seen much difference.

Upvotes: 0

Views: 37

Answers (0)

Related Questions