compiled error about hard-aware-deeply-cascaed-embedding HOT 6 CLOSED

@tendence

(1) My code is compiled with Visual Studio 2013, the refered ‘> >’ may be caused by different platform.

(2)the for (auto const &ent1 : label_data_map) is written in the style of C++11. There should be no problem with this line.

from hard-aware-deeply-cascaed-embedding.

@PkuRainBow
It looks like only when I replaced all your C++11 feature code, can I compile it.

from hard-aware-deeply-cascaed-embedding.

@PkuRainBow
When run the script(has changed to shell), I encountered another problem.
E0213 22:27:34.362665 3195 io.cpp:80] Could not open or find file /home/jack/Downloads/ppdownload/DeepFashion/consumer-to-shop/Img/img/TROUSERS/Pants/id_00017736/comsumer_01.jpg 17734 F0213 22:27:34.362689 3195 image_data_layer.cpp:72] Check failed: cv_img.data Could not load img/TROUSERS/Pants/id_00017736/comsumer_01.jpg 17734
It looks like my io.cpp is different from yours.
It can't seperate the tags from image paths?

from hard-aware-deeply-cascaed-embedding.

@tendence You could check the seperator is "\sp" or "\t". This is a small problem of caffe~ some versions only support "\sp"~

from hard-aware-deeply-cascaed-embedding.

@PkuRainBow Thank you for that.

Besides, is
layer { bottom: "loss1/classifier" top: "loss1/classifier_norm" name: "loss1/l2_norm" type: "Normalization" }
also a specified layer? I found none in the updated BVLC/Caffe repo.

The error message is
Unknown layer type: Normalization

from hard-aware-deeply-cascaed-embedding.

@tendence
Yes, I will commit the L2-norm layer latter today ~

You could refer the implementation normalization_layer.cpp firstly ~

from hard-aware-deeply-cascaed-embedding.