Image Classification using Foundation Models

The aim of this project is to focus on patch-level classification tasks within the domain of whole slide tissue images. The approach involves utilizing pre-trained models as encoders and applying them to the task of patch-level classification. By leveraging pre-trained models, readily available on platforms like Hugging Face, aim to expedite the development process while maintaining high performance. The ultimate objective of the project is to integrate these pre-trained models into caMicroscope, providing users with a convenient tool for downloading and training models tailored to their specific tasks. Deliverables: - A documented method for training a classification model using a pre-trained or foundation model. - Implementation of this method as a set of software components. - Integration of the components with caMicroscope.