I am working on a project for a document classification task by using the image. I have three target classes but the dataset contains images that are out of these classes and marked as others. Any idea on how to approach this without using heavy computational resources?
I tried using a DONUT but its requires very high computational resources.