U-net for image segmentation

162 vues (au cours des 30 derniers jours)
Joseph Stember
Joseph Stember le 22 Août 2017
Commenté : Yi-shan le 18 Juil 2023
I was wondering if it is possible to download U-net (https://lmb.informatik.uni-freiburg.de/people/ronneber/u-net/) for use as a pretrained convolutional neural network for Matlab for medical image segmentation.
  2 commentaires
HARADHAN CHEL le 6 Mar 2020
In matlab documentation, it is clearly written how to build and train a U-net network when the input image and corresponding labelled images are stored into two different folders. But Surprisingly it is not described how to test an image for segmentation on the trained network. For testing images, which command we need to use? is it semanticseg? . Unfortunately this method is not working and not producing any result.

Connectez-vous pour commenter.

Réponses (3)

Moucheng Xu
Moucheng Xu le 16 Août 2018
I think you can look at this one: https://www.mathworks.com/matlabcentral/answers/uploaded_files/107430/createUnet.m
This is a function I found online by mathworks for a modified version U net, I reproduced my own implementation of U net referring to this function so I could make other versions.
  3 commentaires
mohd akmal masud
mohd akmal masud le 12 Août 2021
Modifié(e) : mohd akmal masud le 12 Août 2021
Hi Moucheng Xu,
Is it this code can run in Matlab?

Connectez-vous pour commenter.

Kris Fedorenko
Kris Fedorenko le 25 Août 2017
Hi Joseph,
According to the documentation of u-net, you can download the ready trained network, the source code, the matlab binaries of the modified caffe network, all essential third party libraries and the matlab-interface for overlap-tile segmentation. Best bet would be to use the same setup as recommended by u-net, i.e. Ubuntu Linux 14.04 and Matlab 2014b (x64).
You might also find of interest the image segmentation functionality in the Image Processing Toolbox:
Hope this helps!
  3 commentaires
Yi-shan le 18 Juil 2023
use the image labeler

Connectez-vous pour commenter.

Birju Patel
Birju Patel le 9 Juin 2022
The network you pointed to was trained in Caffe. You can use importCaffeNetwork to import this pretrained U-Net network:


En savoir plus sur Image Data Workflows dans Help Center et File Exchange

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by