Open rickberto opened 1 year ago
Hey, guys, I want to know, I have vertical images in my dataset with height 512 px, and different widths smaller than 512 px, like 512x340, 512x400 etc. Do I have to expand them 512x512 for better training or it doesn't make sense?
No - just enable bucketing, the trainer will take care of the rest.
Hey, guys, I want to know, I have vertical images in my dataset with height 512 px, and different widths smaller than 512 px, like 512x340, 512x400 etc. Do I have to expand them 512x512 for better training or it doesn't make sense?