search

cvlab-stonybrook / SAMPath

Repository for "SAM-Path: A Segment Anything Model for Semantic Segmentation in Digital Pathology" (MedAGI2023, MICCAI2023 workshop)
30 stars 6 forks source link

The preprocess of the dataset #14

Closed zhi-xuan-chen closed 2 months ago

zhi-xuan-chen commented 2 months ago

I want to ask how you preprocess the BCSS and CRAG dataset from the raw dataset?

jingweizhang-xyz commented 2 months ago

For the GlaS dataset, we mainly convert its instance labels to the semantic labels. For the BCSS dataset, we split patches of each file in the dataset as the input resolution of SAM is 1024x1024. Also, we use only tumour, stroma, inflammatory, necrosis and other classes in the dataset.

zhi-xuan-chen commented 2 months ago

How you split them into patches? For example, the overlap between different patches.

jingweizhang-xyz commented 2 months ago

3