Generating Synthetic Image to improve the performance of classifier

I need some suggestion from experts. For my project work, I have been learning about Generative Adversarial Network.

I am trying to make a classifier (say CNN) to train with original CIFAR-10 dataset and then evaluate the performance.

After that, Generating more data of all classes using GAN and retrain the model and compare the performance.

But I don't know how feasible will the work be. How complex will be the work and also I want to know if the idea is acceptable as a project?

1 Answer

I think it could work well, if you try to reduce the false positive error. Generating images for negative examples shouldn't be to difficult. Generating images for positive examples on the other hand would require a good classifier first.