What are the possible ways to handle imbalance in multi-class image datasets?

Question

Image imbalance is one of the major factor in the performance of DL model. Some of the methods that I found to tackle this are oversampling, under-sampling, SMOTE. Over-sampling has cons as it makes model to be overfit.undersampling results in loss of useful information.Again using SMOTE technique also won't works well on image datasets(as per web references).

What is the right way to handle imbalance in image datasets (multi-class problem)?

Answer 1

Yes this is certainly possible...

What you want to do is apply a weight to particular classes by proportion of the imbalance(assuming nothing else related to the problem is of note).

See this post for details