How do you implement class-balanced sampling in terms of 'few' class

Open andotalao24 opened this issue 4 years ago • 1 comments

we define a class where the number of data is less than 20 as few. If some classes only have like 3 pictures. How do you do class-balanced sampling

Apr 17 '21 04:04 andotalao24

maybe you can see https://github.com/facebookresearch/classifier-balancing/issues/9

May 17 '21 12:05 ZSL98