Training GAN on custom dataset

Hi, I have some dataset used for object detection pipeline. I want to use them to produce synthetic images for data augmentation. Is there any idea how to produce dataset pipeline for conditional image generation? And how to train those dataset with labels?

Hi, there are some tools for data augmentation.

E.g. in TensorFlow: Powińôkszanie danych  |  TensorFlow Core

Is that something what you are looking for?

not really. Thanks for the reply :slight_smile: I want to make a custom dataloader for my GAN. The dataset is object detection dataset with coco json file. The GAN will be used for data augmentaion. So, my question is how to design a dataloader to feed the dataset into GAN for training?