error running solve.py image parsing

181 views
Skip to first unread message

Carlos Treviño

unread,
Jul 1, 2015, 7:35:56 AM7/1/15
to caffe...@googlegroups.com
Hi,

has anybody managed to run the fcn-32s- pascal context model froom Model zoo ?

I modified the train_val.prototxt file to run with my training and validation data, but i only get the next message

Process finished with exit code -1073741819 (0xC0000005)

Thanks


sunxu188...@gmail.com

unread,
Jul 1, 2015, 11:26:57 PM7/1/15
to caffe...@googlegroups.com
 I want to reproduce Fully Convolutional Networks for Semantic Segmentation,but I didn't know how to generate  pascal-context-train-lmdb and   pascal-context-train-gt59-lmdb,some days ago ,i just generate lmdb using create.sh in cifar ,but a little difference with this ,looking forword your reply


Carlos Treviño

unread,
Jul 3, 2015, 7:56:26 AM7/3/15
to caffe...@googlegroups.com
Hi,
Here is an example of how to generate the lmdb data needed for image segmentation:

You can prepare the data by making lmdbs in Python: 
https://github.com/BVLC/caffe/issues/1698#issuecomment-70211045. The input images and output images can both be packed into an lmdb in this way. One could also make a PythonLayer to load the data and skip making the lmdbs.
Reply all
Reply to author
Forward
0 new messages