Result is very worse followed by the ipynb file about cascaded-fcn HOT 3 CLOSED

@UpCoder i'm getting the same thing on ubuntu 16.04 with cpu only version. are you able to run it from the docker? with gpu?

from cascaded-fcn.

No, I still not solve the problem. Then I train the model from scratch with tensorflow version. The segmentation of liver is acceptable, but the lesions's can be acceptable.
I don't run it from the docker, I directly use the GPU server.

from cascaded-fcn.

thanks @mohamed-ezz

seems that you need to use the same caffe version from the docker:

root@maeda:/opt/caffe# cat .git/config
[core]
repositoryformatversion = 0
filemode = true
bare = false
logallrefupdates = true
[remote "origin"]
url = https://github.com/mohamed-ezz/caffe.git
fetch = +refs/heads/:refs/remotes/origin/
[branch "master"]
remote = origin
merge = refs/heads/master
[branch "jonlong"]
remote = origin
merge = refs/heads/jonlong
root@maeda:/opt/caffe# git log
commit 876e387a6d7f8974f68f42beacd3728b4fc92ff7
Author: Mohamed Ezz [email protected]
Date: Thu Feb 18 01:24:04 2016 +0100

Add class weighting feature for softmax_loss layer

$ git clone https://github.com/mohamed-ezz/caffe.git caffe/docker
$ cd caffe/docker; git branch docker 876e387a6d7f8974f68f42beacd3728b4fc92ff7 ; git checkout docker

from cascaded-fcn.