search

junyanz / CycleGAN

Software that can generate photos from paintings, turn horses into zebras, perform style transfer, and more.
Other
12.22k stars 1.94k forks source link

The number of training dataset #114

Open qianxiao111 opened 5 years ago

qianxiao111 commented 5 years ago

Hello, I have a question. How many images in the training dataset can be used to train an effective model? For example, domain X has 6000 undistorted images, domain Y has only 145 images with five distortion types. The scenes of Y are included in X . In this case, could generate images with five distortion types, which is similar to Y domain?

junyanz commented 5 years ago

I will recommend 1000+ images per dataset. But you may want to try your dataset and see how it goes.

qianxiao111 commented 5 years ago

Thanks, I have a question, if the Y domain constain five distortion types with one distortion level, is the generated images mixed distortions with five distortion types or single distortion?

junyanz commented 5 years ago

probably multiple distortions.