SKINET Project is meant to provide a segmentation of the different structures in kidney histological tissue (biopsy or nephrectomy). It allows the segmentation of sclerotic and non sclerotic glomeruli, healthy or atrophic tubules, veins... This repository contains all the project's code. You can use our online inference tool to test our tool on your biopsies (tutorial in the "docs" folder). You can also create a local version of this project by cloning this repo and installing a suitable environment (tutorial in the "docs" folder).

The project's code is based on Matterport's Mask R-CNN and Navidyou's repository.

This project is a collaboration between a Nephrology team from Dijon Burgundy Teaching Hospital, LEAD Laboratory, and a student from ESIREM, all located in Dijon, Burgundy, France.

Last :

v1.0 :

v1.1.1 :

Update: Unfortunately, google collab doesn't support tensorflow 1* anymore. The online version of our tool is no longer available. We are currently working on providing a new version. In the mean time, you can still clone this repo and use a local version.