DeePathology is proud to bring its Unsupervised Semantic Segmentation for Pathology by Factorizing Foundation Model Features as an open source!
Based on our published a method called “Segmentation by Factorization“, that allows utilizing foundational models for unsupervised segmentation, AS IS, WITHOUT any training.
This unlocks many possibilities:
🧠 Utilize pathology foundational models for unsupervised semantic segmentation.
🏷️ No need for any annotated data!
🚀 Generates powerful H&E image segmentation, for a configurable number of semantic classes.
We have implemented these amazing capabilities within our DeePathology® STUDIO and are also committed to open source.
So, we are proud to share our Unsupervised Semantic Segmentation for Pathology in GitHub
