UNI is a large self-supervised vision transformer trained on diverse histology images for feature extraction in computational pathology. Developed by Mahmood Lab at Harvard, it is released under a non-commercial research license and integrates with the timm library. It enables researchers to build accurate models for cancer detection, biomarker prediction, and other medical imaging tasks using transfer learning.
In the Other AI space, UNI takes a focused approach. It focuses on extracting rich, generalizable features from gigapixel histology images for downstream pathology AI tasks without large labeled datasets. It is built as an open-source project for researchers. UNI is open source under the Open Source license. It runs on the web and API.
MahmoodLab builds and maintains UNI, and it first shipped in 2024. The project is developed in the open on GitHub with 763 stars. Key capabilities include Feature Extraction, histopathology, and Self-supervised Learning.
Summary written by a language model from the project’s public pages.
What PulseGate has recorded for this listing
Closest matches by what these projects do