hyeonwoonoh/caffe

modified version of caffe which support DeconvNet and DecoupledNet

caffe's Issues

Tranformer attribute missing

Cannot find the function Transformer to preprocess the image.

File "tst.py", line 18, in
transformer = caffe.io.Transformer({'data': net.blobs['data'].data.shape})
AttributeError: 'module' object has no attribute 'Transformer

No CaffeConfig.cmake. Can't configured on Ubuntu without root access.

CMake Warning at CMakeLists.txt:33 (find_package):
By not providing "FindCaffe.cmake" in CMAKE_MODULE_PATH this project has
asked CMake to find a package configuration file provided by "Caffe", but
CMake did not find one.

Could not find a package configuration file provided by "Caffe" with any of
the following names:

 CaffeConfig.cmake
 caffe-config.cmake

Add the installation prefix of "Caffe" to CMAKE_PREFIX_PATH or set
"Caffe_DIR" to a directory containing one of the above files. If "Caffe"
provides a separate development package or SDK, be sure it has been
installed.

Erros in 'make all'

I'm try to build (make all) this code with Ubuntu 14.04, cuda-7.5, cudnn-v4, Titan X.
But, I got belows messages.
In my case, this code was only success without USE_CUDNN.
Also, There was no problem with the newest original code of caffe.
Doesn't this code work with USE_CUDNN?
Please help me.

sm5@sv:~/caffe-master$ make all
PROTOC src/caffe/proto/caffe.proto
PROTOC src/caffe/proto/caffe_pretty_print.proto
CXX .build_release/src/caffe/proto/caffe.pb.cc
CXX .build_release/src/caffe/proto/caffe_pretty_print.pb.cc
CXX src/caffe/solver.cpp
In file included from ./include/caffe/util/device_alternate.hpp:40:0,
from ./include/caffe/common.hpp:18,
from ./include/caffe/blob.hpp:4,
from ./include/caffe/net.hpp:10,
from src/caffe/solver.cpp:7:
./include/caffe/util/cudnn.hpp:60:32: error: variable or field ‘createTensor4dDesc’ declared void
inline void createTensor4dDesc(cudnnTensor4dDescriptor_t* desc) {
^
./include/caffe/util/cudnn.hpp:60:32: error: ‘cudnnTensor4dDescriptor_t’ was not declared in this scope
./include/caffe/util/cudnn.hpp:60:59: error: ‘desc’ was not declared in this scope

...
...
...

./include/caffe/util/cudnn.hpp:113:13: error: ‘CUDNN_POOLING_AVERAGE’ was not declared in this scope
*mode = CUDNN_POOLING_AVERAGE;
^
./include/caffe/util/cudnn.hpp:120:27: error: there are no arguments to ‘cudnnSetPoolingDescriptor’ that depend on a template parameter, so a declaration of ‘cudnnSetPoolingDescriptor’ must be available [-fpermissive]
stride_h, stride_w));
^
./include/caffe/util/cudnn.hpp:12:28: note: in definition of macro ‘CUDNN_CHECK’
cudnnStatus_t status = condition; \

Can't install your caffe code, make errors

Ubuntu 14.04, OpenCV 3 , cuda 7.0
Everything goes well with original branch of Caffe, but i can't compile your Caffe code, with or without comment "USE_CUDNN := 1", i don't know why, is the version of OpenCV doesn't match? or something else?
I got an error like below, with comment "USE_CUDNN := 1"

Any suggestion? Thanks

cudnn error

when i build the project with cudnn enable .it gives me a error .project has used a struct that does not initialized in cuddn.h .i checked it. cudnntensor4dstruct is the struct that i memtion.so what to do to solve this problem

Ubuntu14.04 server x64, OpenCV3, CuDNNv3, make errors.

Errors like "./include/caffe/util/cudnn.hpp:73:29: error: ‘cudnnTensor4dDescriptor_t’ was not declared in this scope" appears.
After I comment 'USE CuDNN' in Makefile.config, I meet these:

"
src/caffe/layers/window_data_layer.cpp:26:11: error: ‘const int CV_LOAD_IMAGE_COLOR’ redeclared as different kind of symbol
const int CV_LOAD_IMAGE_COLOR = cv::IMREAD_COLOR;
^
In file included from /usr/local/include/opencv2/highgui/highgui_c.h:47:0,
from /usr/local/include/opencv2/highgui.hpp:701,
from /usr/local/include/opencv2/highgui/highgui.hpp:48,
from src/caffe/layers/window_data_layer.cpp:10:
/usr/local/include/opencv2/imgcodecs/imgcodecs_c.h:62:5: error: previous declaration of ‘ CV_LOAD_IMAGE_COLOR’
CV_LOAD_IMAGE_COLOR =1,
^
make: *** [.build_release/src/caffe/layers/window_data_layer.o] Error 1
make: *** Waiting for unfinished jobs....
"
and some other warnings about opencv.

I've already deployed the newest version of caffe, so I think my opencv and cudnn are all right.

Thanks.

Installation Error (Ubuntu 14.04 - Remote machine with No Root Access)

When I try to set up modified Caffe Version to execute DeConvnet, I get errors in Make all as:

CXX/LD -o .build_release/test/test_all.testbin src/caffe/test/test_caffe_main.cpp
/usr/lib/gcc/x86_64-linux-gnu/4.8/../../../x86_64-linux-gnu/libopencv_highgui.so: undefined reference to TIFFReadRGBAStrip@LIBTIFF_4.0' /usr/lib/gcc/x86_64-linux-gnu/4.8/../../../x86_64-linux-gnu/libopencv_highgui.so: undefined reference toTIFFIsTiled@LIBTIFF_4.0'
//usr/lib/x86_64-linux-gnu/libunwind.so.8: undefined reference to lzma_stream_buffer_decode@XZ_5.0' /usr/lib/gcc/x86_64-linux-gnu/4.8/../../../x86_64-linux-gnu/libopencv_highgui.so: undefined reference toTIFFWriteScanline@LIBTIFF_4.0'

---

I Would like to know which version of OpenCV is needed for this modified Caffe also Is it mandatory to install Opencv to get this running ? Kindly Help !! Thank you.

Multinomial Logistical Loss

Is there any chance to use MultinomialLogisticalLossLayer for this architecture?

My problem relies on I have tiny objects to classify in each image and using softmaxlayer, the network always "learn" to predict background label, which indeed is around 95% of the image. My problem does not admit edge-boxes extracting object proposals. Someone advice me to use logistic loss.

Net-surgery, python

I am trying to perform net surgery, in order to get the weigths from another pre-trained model,) but when I type this in python I get an error:
net = caffe.Net(model, weight, caffe.TEST)
AttributeError: 'module' object has no attribute 'TEST'

The same thing for caffe.TRAIN

If I type 1 or 0, which means those before, I get this error:

ArgumentError: Python argument types in
Net.init(Net, str, str, int)
did not match C++ signature:
init(object, std::string)
init(object, std::string, std::string)

Do you know how to fix this?

which files in your version are changed respect to the version in the BVLC

Can you tell me which files in your version are changed respect to the version in the BVLC, thanks!

Matcaffe

Hi,
I was wondering is this version of caffe can be interfaced with MATLAB or not? When I tried to make matcaffe I faced an error since there is no +caffe folder in /matlab directory.

Thanks!
best
Ara

Link Error

I‘ve follow the suggestions #9 ’ that fix this problem by using cuDNN v1 cudnn.h file. However, I get the following errors.

Any suggestions as to what might be causing this?

Problem with installation

Hi,

I'm noticing the following errors while trying to install your version of caffe:
$ make all-j8
Normal output barring this warning:
"src/caffe/layers/unpooling_layer.cu(100): warning: variable "pool_size" was declared but never referenced
detected during instantiation of "void caffe::UnpoolingLayer::Forward_gpu(const std::vectorcaffe::Blob<Dtype *, std::allocatorcaffe::Blob<Dtype *>> &, const std::vectorcaffe::Blob<Dtype *, std::allocatorcaffe::Blob<Dtype *>> &) [with Dtype=float]"
(290): here

src/caffe/layers/unpooling_layer.cu(100): warning: variable "pool_size" was declared but never referenced
detected during instantiation of "void caffe::UnpoolingLayer::Forward_gpu(const std::vectorcaffe::Blob<Dtype *, std::allocatorcaffe::Blob<Dtype *>> &, const std::vectorcaffe::Blob<Dtype *, std::allocatorcaffe::Blob<Dtype *>> &) [with Dtype=float]"
(290): here

src/caffe/layers/unpooling_layer.cu(100): warning: variable "pool_size" was declared but never referenced
detected during instantiation of "void caffe::UnpoolingLayer::Forward_gpu(const std::vectorcaffe::Blob<Dtype *, std::allocatorcaffe::Blob<Dtype *>> &, const std::vectorcaffe::Blob<Dtype *, std::allocatorcaffe::Blob<Dtype *>> &) [with Dtype=float]"
(290): here

src/caffe/layers/unpooling_layer.cu(100): warning: variable "pool_size" was declared but never referenced

src/caffe/layers/unpooling_layer.cu(100): warning: variable "pool_size" was declared but never referenced

src/caffe/layers/unpooling_layer.cu(100): warning: variable "pool_size" was declared but never referenced"

$ make test
Normal

$ make runtest
"src/caffe/test/test_math_functions.cpp:202: Failure
Value of: x[i] > 0 ? x[i] : -x[i]
Actual: 0.233989
Expected: abs_val[i]
Which is: 0"
[ FAILED ] MathFunctionsTest/1.TestFabsGPU, where TypeParam = double (33444 ms)

Kindly let me know whats wrong here. Any help will be appreciated.

When I try to set up modified Caffe Version to execute DeConvnet, I get errors in 'make install' as:

CMake Error at src/caffe/proto/cmake_install.cmake:40 (FILE):
file INSTALL cannot find
"/home/will/deepLearning/DecoupledNet/caffe-master/build/src/caffe/proto/caffe_pretty_print_pb2.py".
Call Stack (most recent call first):
src/caffe/cmake_install.cmake:41 (INCLUDE)
cmake_install.cmake:42 (INCLUDE)

make: *** [install] Error 1

I don't find caffe_pretty_print_pb2.py in the modified Caffe Vision, and caffe_pb2.py too, where to get them? Kindly Help !! Thank you.

I am having cudasuccess error

I am trying to do convolutional autoencoders and I've got this error: cudaSuccess (8 vs. 0) invalid device function

my card is gtx 960 and it has cuda capabilities. do u know why?

fast FwdMaxUnpooling GPU kernel

in file https://github.com/HyeonwooNoh/caffe/blob/master/src/caffe/layers/unpooling_layer.cpp
you asked: TODO(Yangqing): Is there a faster way to do unpooling in the channel-first

I tried also to implement a fast ForwardMaxUnPooling.
I only made the GPU version. See at the bottom of this file:
https://github.com/3DdeepAI/3DdeepAI/blob/3DdeepAI/NNkernels.h
kernel "FwdMaxUnpooling"

In my implementation there are less divisions: 2 instead of 6

https://github.com/HyeonwooNoh/caffe/blob/master/src/caffe/layers/unpooling_layer.cu

Unable to build with cuDNN flag

I've tried using cudnn versions 2, 3 & 4, none of which work.
I have no problems making on the master caffe with USE_CUDNN := 1, however, when I try to make this branch of caffe, I get the following errors:

make: *** [.build_release/src/caffe/lmdb_dataset.o] Error 1
In file included from ./include/caffe/util/device_alternate.hpp:40:0,
from ./include/caffe/common.hpp:18,
from ./include/caffe/internal_thread.hpp:4,
from src/caffe/internal_thread.cpp:1:
./include/caffe/util/cudnn.hpp:60:32: error: variable or field ‘createTensor4dDesc’ declared void
inline void createTensor4dDesc(cudnnTensor4dDescriptor_t* desc) {
^
./include/caffe/util/cudnn.hpp:60:32: error: ‘cudnnTensor4dDescriptor_t’ was not declared in this scope
./include/caffe/util/cudnn.hpp:60:59: error: ‘desc’ was not declared in this scope
inline void createTensor4dDesc(cudnnTensor4dDescriptor_t* desc) {
^
./include/caffe/util/cudnn.hpp:65:29: error: variable or field ‘setTensor4dDesc’ declared void
inline void setTensor4dDesc(cudnnTensor4dDescriptor_t* desc,
^
./include/caffe/util/cudnn.hpp:65:29: error: ‘cudnnTensor4dDescriptor_t’ was not declared in this scope
./include/caffe/util/cudnn.hpp:65:56: error: ‘desc’ was not declared in this scope
inline void setTensor4dDesc(cudnnTensor4dDescriptor_t* desc,
^
... etc.

Any suggestions as to what might be causing this?