Problem with load_image

[kamenix]

Hello,

I tried to run the code. I had a problem with this line :

[source_img, net] = load_image(im_dir + source_img_name, im_size, VGGmodel, VGGweights, imagenet_mean, show_img=True). The program dies and here is an extract from the console :

5717 net.cpp:219] conv3_2 does not need backward computation.
I0301 15:29:31.425060 5717 net.cpp:219] relu3_1 does not need backward computation.
I0301 15:29:31.425073 5717 net.cpp:219] conv3_1 does not need backward computation.
I0301 15:29:31.425088 5717 net.cpp:219] pool2 does not need backward computation.
I0301 15:29:31.425102 5717 net.cpp:219] relu2_2 does not need backward computation.
I0301 15:29:31.425115 5717 net.cpp:219] conv2_2 does not need backward computation.
I0301 15:29:31.425130 5717 net.cpp:219] relu2_1 does not need backward computation.
I0301 15:29:31.425143 5717 net.cpp:219] conv2_1 does not need backward computation.
I0301 15:29:31.425158 5717 net.cpp:219] pool1 does not need backward computation.
I0301 15:29:31.425173 5717 net.cpp:219] relu1_2 does not need backward computation.
I0301 15:29:31.425186 5717 net.cpp:219] conv1_2 does not need backward computation.
I0301 15:29:31.425202 5717 net.cpp:219] relu1_1 does not need backward computation.
I0301 15:29:31.425215 5717 net.cpp:219] conv1_1 does not need backward computation.
I0301 15:29:31.425230 5717 net.cpp:219] input does not need backward computation.
I0301 15:29:31.425252 5717 net.cpp:261] This network produces output pool5
I0301 15:29:31.425343 5717 net.cpp:274] Network initialization done.
I0301 15:29:32.167654 5717 upgrade_proto.cpp:51] Attempting to upgrade input file specified using deprecated V1LayerParameter: /home/mkammoun/PycharmProjects/Deep_Texture_2/Models/VGG_normalised.caffemodel
I0301 15:29:32.502588 5717 upgrade_proto.cpp:59] Successfully upgraded file specified using deprecated V1LayerParameter
I0301 15:29:32.506871 5717 upgrade_proto.cpp:65] Attempting to upgrade input file specified using deprecated input fields: /home/mkammoun/PycharmProjects/Deep_Texture_2/Models/VGG_normalised.caffemodel

Process finished with exit code 139

(I am using Pycharm)

After debugging, it comes to a problem of loading the pretrained network inside the function load_image :

net = caffe.Classifier(
net_model, net_weights,
mean = net_mean,
channel_swap=(2,1,0),
input_scale=255,)

There is not much on google, anyone can help ? thanks

[leongatys]

Hi, I am sorry but I couldn't reproduce your problem. Could it be that there is a problem with your caffe version (the syntax of loading models changed between different caffe versions)?
I definitely works for me using the Caffe 0.14 Nvidia binaries.
You can get a docker container with my setup by following the instructions here: https://github.com/bethgelab/docker

Best
Leon