0

Can the offset dataset handling method using the SMOTE (Synthetic Minority Oversampling Technique) method be applied to image datasets? because as far as I know SMOTE is only used for structured data, if the SMOTE method can be applied to image datasets, friends, please refer to the source code related to that.

please help me, thanks you

Anju Ucok Lubis
  • 11
  • 2
  • what have you tried so far ? the question needs sufficient code for a minimal reproducible example: https://stackoverflow.com/help/minimal-reproducible-example – D.L Feb 23 '23 at 14:14
  • What would you like to achieve with SMOTE on images? Might stable diffusion help for your use case (e.g. generating more image samples for training)? – Ken Jiiii Apr 17 '23 at 12:59
  • @KenJiiii the number of images that I want to train is different for each class, I want to balance the number of images using SMOTE, previously it was wrong to do an experiment using augmentation, but I want to try an experiment to handle image dataset imbalances using SMOTE. – Anju Ucok Lubis Aug 01 '23 at 04:09
  • SMOTE is a type of data augmentation. Maybe checkout https://medium.com/swlh/how-to-use-smote-for-dealing-with-imbalanced-image-dataset-for-solving-classification-problems-3aba7d2b9cad and https://stackoverflow.com/questions/53666759/use-smote-to-oversample-image-data – Ken Jiiii Aug 01 '23 at 12:14

0 Answers0