The aim of the project is to study the possibility to combine two image processing paradigm; one coming from the classical image processing techniques:
-
the super-pixels and
-
the other belong to the more recent deep learning network .
The development will benefit from the GPU LISA infrastructure.
Contacts:
Olivier Debeir odebeir@ulb.ac.be
Dear Pr.Debeir,
I forgot to tell you that I found a database for my project ( mix super pixel and deep learning for segmentation).
It is a database of Ultrasound image to segmentize nerve. You may find examples on this mail.
Here is the link for this database: https://www.kaggle.com/c/ultrasound-nerve-segmentation/overview/description
Moreover, since the university became in “code orange”, will the reunion of Monday 10h be remotely? Because we do not have a group on teams for this project.
Kind regards,
superpixels definition
ultrasound
segment the nerve
apply on US not maybe good idea to be tested
superpixels skimage
on laisse tomber les SP pour le problème US nerfs
test avec librairie (backbones) mais problème d'import
pour de la segmentation il ne faut pas utiliser un dense à la fin !
repartir depuis un exemple unet ou segnet
un peu à la bourre, se demande si pertinent de reporter l'année prochaine