rbgirshick / caffe-fast-rcnn Goto Github PK

Hi. I am trying to make caffe and I am getting following error. Can anyone help me to solve the problem? I have attached my Makefile.config file.

/usr/bin/ld: /usr/local/lib/libleveldb.a(db_impl.cc.o): relocation R_X86_64_32 against `_ZTVN7leveldb6DBImplE' can not be used when making a shared object; recompile with -fPIC
/usr/local/lib/libleveldb.a: error adding symbols: Bad value
collect2: error: ld returned 1 exit status
Makefile:582: recipe for target '.build_release/lib/libcaffe.so.1.0.0' failed
make: *** [.build_release/lib/libcaffe.so.1.0.0] Error 1

I have tried both configs and I get same error:

Operating system: ubuntu 16.04
CUDA version:10.2
CUDNN version: 7

and

Operating system: ubuntu 16.04
CUDA version:8
CUDNN version: 5

Makefile.config.zip

Hello,

So I am a new faster-rcnn user and I am trying to build caffe as CPU-only mode on my mac (10.11.6) . I am able to do make, make test, and make runtest by commenting out some GPU related code and modifying the Makefile.config file. However, when I open python and do import caffe I get the following error:

Python 2.7.9 |Anaconda custom (x86_64)| (default, Dec 15 2014, 10:37:34)
[GCC 4.2.1 (Apple Inc. build 5577)] on darwin
Type "help", "copyright", "credits" or "license" for more information.
Anaconda is brought to you by Continuum Analytics.
Please check out: http://continuum.io/thanks and https://binstar.org

import caffe
Traceback (most recent call last):
File "", line 1, in
File "/Users/Sirius_zn/Desktop/C3PO/full_installation_python/py-faster-rcnn/caffe-fast-rcnn/python/caffe/init.py", line 1, in
from .pycaffe import Net, SGDSolver, NesterovSolver, AdaGradSolver, RMSPropSolver, AdaDeltaSolver, AdamSolver
File "/Users/Sirius_zn/Desktop/C3PO/full_installation_python/py-faster-rcnn/caffe-fast-rcnn/python/caffe/pycaffe.py", line 13, in
from ._caffe import Net, SGDSolver, NesterovSolver, AdaGradSolver,
ImportError: dlopen(/Users/Sirius_zn/Desktop/C3PO/full_installation_python/py-faster-rcnn/caffe-fast-rcnn/python/caffe/_caffe.so, 2): Symbol not found: ___addtf3
Referenced from: /usr/local/opt/gcc/lib/gcc/6/libquadmath.0.dylib
Expected in: /usr/local/lib/libgcc_s.1.dylib
in /usr/local/opt/gcc/lib/gcc/6/libquadmath.0.dylib

I have double checked that I have _caffe.so in the path mentioned above.

Sirius_zn (master *) tools $ ls /Users/Sirius_zn/Desktop/C3PO/full_installation_python/py-faster-rcnn/tools/../caffe-fast-rcnn/python/caffe/
init.py _caffe.so draw.py net_spec.py pycaffe.pyc
init.pyc classifier.py imagenet proto test
_caffe.cpp detector.py io.py pycaffe.py

I have been googling around for a while, and could not bypass this error. I tried setting PYTHONPATH.

Sirius_zn (master *) tools $ echo $PYTHONPATH
/Users/Sirius_zn/Desktop/C3PO/full_installation_python/py-faster-rcnn/caffe-fast-rcnn/python/

Also I have seen some issue about mac's System Integrity Protection might be causing the problem.

Can anyone help?

Thanks!!

Thank you very much @rbgirshick

as title

Hi,

I am wondering why need propagate back to bottom[1] in Smooth_L1_Loss? Is it strange to propagate to target box which actually is label for bbox regression?

Thanks much!

This is in the following code:
for (int i = 0; i < 2; ++i) {
if (propagate_down[i]) {
const Dtype sign = (i == 0) ? 1 : -1;
const Dtype alpha = sign * top[0]->cpu_diff()[0] / bottom[i]->num();
caffe_gpu_axpby(
bottom[i]->count(), // count
alpha, // alpha
diff_.gpu_data(), // x
Dtype(0), // beta
bottom[i]->mutable_gpu_diff()); // y
}
}
}

i am under mac os 10.12.3 Sierra .
when i run : hamdihamed$ python /Users/hamdihamed/Desktop/py-faster-rcnn/tools/demo.py
i get this error
Traceback (most recent call last):
File "/Users/hamdihamed/Desktop/py-faster-rcnn/tools/demo.py", line 18, in
from fast_rcnn.test import im_detect
File "/Users/hamdihamed/Desktop/py-faster-rcnn/tools/../lib/fast_rcnn/test.py", line 16, in
import caffe
File "/Users/hamdihamed/Desktop/py-faster-rcnn/tools/../caffe-fast-rcnn/python/caffe/init.py", line 1, in
from .pycaffe import Net, SGDSolver, NesterovSolver, AdaGradSolver, RMSPropSolver, AdaDeltaSolver, AdamSolver
File "/Users/hamdihamed/Desktop/py-faster-rcnn/tools/../caffe-fast-rcnn/python/caffe/pycaffe.py", line 13, in
from ._caffe import Net, SGDSolver, NesterovSolver, AdaGradSolver,
ImportError: dlopen(/Users/hamdihamed/Desktop/py-faster-rcnn/tools/../caffe-fast-rcnn/python/caffe/_caffe.so, 2): Library not loaded: libcaffe.so.1.0.0-rc3
Referenced from: /Users/hamdihamed/Desktop/py-faster-rcnn/caffe-fast-rcnn/python/caffe/_caffe.so
Reason: image not found

any help ??

Hi,
Say i have a region proposal with co-ordinates (top left, bottom right coordinates): (0, 3), (7, 8) and the output i am taking out after roi pooling is 2x2. So, according to the roi pooling code in this repo, we create 4 regions of size([height,width]) [floor(7/2), floor(5/2)],i.e,[3,2]. But in this case, the four regions of size [3,2] would cover only [6,4] area of the region proposal and the features from the rest of the proposal gets ignored.

I came across this blog: https://deepsense.ai/region-of-interest-pooling-explained/ where they select regions like below

Here, the first box has size [3,2] while the rest of the boxes inside the proposal regions have their sizes adjusted so as to cover and extract features from the entire proposal region.

So, wouldnt the approach mentioned in the above blog be better or is that somehow being ensured in your implementation?

Specifically I am looking for a way to get bounding boxes for the proposed region and their classification.

...
...
I0413 14:52:27.871963 14366 layer_factory.hpp:77] Creating layer rpn_cls_prob_reshape
I0413 14:52:27.871973 14366 net.cpp:106] Creating Layer rpn_cls_prob_reshape
I0413 14:52:27.871980 14366 net.cpp:454] rpn_cls_prob_reshape <- rpn_cls_prob
I0413 14:52:27.871995 14366 net.cpp:411] rpn_cls_prob_reshape -> rpn_cls_prob_reshape
I0413 14:52:27.872011 14366 net.cpp:150] Setting up rpn_cls_prob_reshape
I0413 14:52:27.872020 14366 net.cpp:157] Top shape: 1 18 14 14 (3528)
I0413 14:52:27.872026 14366 net.cpp:165] Memory required for data: 233788800
I0413 14:52:27.872032 14366 layer_factory.hpp:77] Creating layer proposal
ImportError: No module named rpn.proposal_layer
terminate called after throwing an instance of 'boost::python::error_already_set'
The program has unexpectedly finished.
/home/hami/INOBOX/Caffe_Projects/caffeProject/build-caffeClassifier-Desktop-Debug/caffeClassifier crashed.

@rbgirshick I saw you update the caffe of faster rcnn to the 33f2445 version. What's the difference of this caffe to https://github.com/ShaoqingRen/caffe/tree/faster-R-CNN?

Since I cannot compare these two repo, can you suggest the major change needed if I want to update the caffe for matlab version of faster-rcnn ?

Hi, I was confused about the smooth_l1_loss_layer.cu， that the diff_ in Forward_gpu has multiply the "inside" weights, why in Backward_gpu, multiply the "inside" weights again?

As title

Hi,
when I'm compiling the caffe-fast-rcnn, these two steps are right:
make all -j8
make test -j8
but the error happens when I "make runtest -j8" , and the error information is as follows:

[----------] 6 tests from CuDNNConvolutionLayerTest/1, where TypeParam = double
[ RUN ] CuDNNConvolutionLayerTest/1.TestGradientCuDNN
*** Aborted at 1521030707 (unix time) try "date -d @1521030707" if you are using GNU date ***
PC: @ 0x7fe3c3775512 cfree
*** SIGSEGV (@0x9) received by PID 19280 (TID 0x7fe3cd080740) from PID 9; stack trace: ***
@ 0x7fe3c3acc390 (unknown)
@ 0x7fe3c3775512 cfree
@ 0x7fe3c450878c __gnu_cxx::new_allocator<>::deallocate()
@ 0x7fe3c450853e std::allocator_traits<>::deallocate()
@ 0x7fe3c4507ffe std::_Vector_base<>::_M_deallocate()
@ 0x7fe3c450793b std::_Vector_base<>::~_Vector_base()
@ 0x7fe3c45068e5 std::vector<>::~vector()
@ 0x7fe3c45067c6 caffe::CuDNNConvolutionLayer<>::~CuDNNConvolutionLayer()
@ 0x51ca79 caffe::CuDNNConvolutionLayerTest_TestGradientCuDNN_Test<>::TestBody()
@ 0x83f704 testing::internal::HandleSehExceptionsInMethodIfSupported<>()
@ 0x83ac91 testing::internal::HandleExceptionsInMethodIfSupported<>()
@ 0x826c82 testing::Test::Run()
@ 0x8273fe testing::TestInfo::Run()
@ 0x8279b9 testing::TestCase::Run()
@ 0x82cccb testing::internal::UnitTestImpl::RunAllTests()
@ 0x8405f7 testing::internal::HandleSehExceptionsInMethodIfSupported<>()
@ 0x83b99a testing::internal::HandleExceptionsInMethodIfSupported<>()
@ 0x82ba2c testing::UnitTest::Run()
@ 0x48f8e4 main
@ 0x7fe3c3711830 __libc_start_main
@ 0x48f6c9 _start
@ 0x0 (unknown)
Makefile:515: recipe for target 'runtest' failed
make: *** [runtest] 段错误 (core dumped)

What is the reason?

Hi, I have the problem that how to set multiple gpu devices with python interface, like caffe.set_device() .

I use CUDA 8.0 and cuDNN v5. I cannot make the project for some issues of API inconsistency like:

CXX src/caffe/util/cudnn.cpp
In file included from src/caffe/util/cudnn.cpp:2:0:
./include/caffe/util/cudnn.hpp: In function ‘void caffe::cudnn::createPoolingDesc(cudnnPoolingStruct**, caffe::PoolingParameter_PoolMethod, cudnnPoolingMode_t*, int, int, int, int, int, int)’:
./include/caffe/util/cudnn.hpp:124:41: error: too few arguments to function ‘cudnnStatus_t cudnnSetPooling2dDescriptor(cudnnPoolingDescriptor_t, cudnnPoolingMode_t, cudnnNanPropagation_t, int, int, int, int, int, int)’
pad_h, pad_w, stride_h, stride_w));
^
./include/caffe/util/cudnn.hpp:12:28: note: in definition of macro ‘CUDNN_CHECK’
cudnnStatus_t status = condition;
^
In file included from ./include/caffe/util/cudnn.hpp:5:0,
from src/caffe/util/cudnn.cpp:2:
/usr/local/cuda/include/cudnn.h:799:27: note: declared here
cudnnStatus_t CUDNNWINAPI cudnnSetPooling2dDescriptor(
^
Makefile:501: recipe for target '.build_release/src/caffe/util/cudnn.o' failed
make: *** [.build_release/src/caffe/util/cudnn.o] Error 1

I am a green hand and sorry to bother you about this closed project, but I'm quite confused about the test.prototxt of the VGG16 model, in which the input is
input: "data"
input_shape {
dim: 1
dim: 3
dim: 224
dim: 224
}

input: "rois"
input_shape {
dim: 1 # to be changed on-the-fly to num ROIs
dim: 5 # [batch ind, x1, y1, x2, y2] zero-based indexing
}

is this natively supported by caffe or Mr. RGB change the source code?

Hi @rbgirshick,

I've read your RoI pooling implementation and found it very interesting that pooling regions in one RoI overlap.

Let's say that we have one one channel feature map 5x5 and that we have only one RoI (0, 0, 4, 4) with pooled_height_ and pooled_width_ set to 2. Current implementation makes 4 pools in 4 regions and each region is of size 3x3 and the common element of those 4 pool regions is element (2,2).

Is this expected behaviour? Should pooled regions overlap? Why is that prefered? I've read this article where thats not the case and that's why I am confused with your implementation.

BR,
Herman

I wanted to apply Faster RCNN on oriented bounding box as you know each oriented bounding box has 8 coordinates, and I dont know how I would train with these 8 coordinates. Can you guide me?

Guys please help me how to deal with?
I check all people solutions from thread of same issue! nothing stop this error.

UBUNTU 17.10
CUDA 8
CUDNN 5
gcc (Ubuntu 4.8.5-4ubuntu6) 4.8.5
protoc --version: libprotoc 3.5.1

I use this caffe version which support CUDNN5:
sudo git clone https://github.com/nils489/caffe-fast-rcnn.git

[ 80%] Linking CXX executable extract_features
CMakeFiles/extract_features.dir/extract_features.cpp.o: In function `std::string* google::MakeCheckOpString<int, int>(int const&, int const&, char const*)':
extract_features.cpp:(.text._ZN6google17MakeCheckOpStringIiiEEPSsRKT_RKT0_PKc[_ZN6google17MakeCheckOpStringIiiEEPSsRKT_RKT0_PKc]+0x43): undefined reference to `google::base::CheckOpMessageBuilder::NewString()'
CMakeFiles/extract_features.dir/extract_features.cpp.o: In function `std::string* google::MakeCheckOpString<unsigned int, int>(unsigned int const&, int const&, char const*)':
extract_features.cpp:(.text._ZN6google17MakeCheckOpStringIjiEEPSsRKT_RKT0_PKc[_ZN6google17MakeCheckOpStringIjiEEPSsRKT_RKT0_PKc]+0x43): undefined reference to `google::base::CheckOpMessageBuilder::NewString()'
CMakeFiles/extract_features.dir/extract_features.cpp.o: In function `std::string* google::MakeCheckOpString<unsigned long, unsigned long>(unsigned long const&, unsigned long const&, char const*)':
extract_features.cpp:(.text._ZN6google17MakeCheckOpStringImmEEPSsRKT_RKT0_PKc[_ZN6google17MakeCheckOpStringImmEEPSsRKT_RKT0_PKc]+0x44): undefined reference to `google::base::CheckOpMessageBuilder::NewString()'
CMakeFiles/extract_features.dir/extract_features.cpp.o: In function `int feature_extraction_pipeline<float>(int, char**)':
extract_features.cpp:(.text._Z27feature_extraction_pipelineIfEiiPPc[_Z27feature_extraction_pipelineIfEiiPPc]+0xd65): undefined reference to `google::protobuf::internal::empty_string_'
extract_features.cpp:(.text._Z27feature_extraction_pipelineIfEiiPPc[_Z27feature_extraction_pipelineIfEiiPPc]+0xf3d): undefined reference to `google::protobuf::MessageLite::SerializeToString(std::string*) const'
../lib/libcaffe.so: undefined reference to `google::protobuf::Message::InitializationErrorString() const'
../lib/libcaffe.so: undefined reference to `google::protobuf::internal::WireFormatLite::WriteStringMaybeAliased(int, std::string const&, google::protobuf::io::CodedOutputStream*)'
../lib/libcaffe.so: undefined reference to `google::protobuf::io::CodedOutputStream::WriteStringWithSizeToArray(std::string const&, unsigned char*)'
../lib/libcaffe.so: undefined reference to `google::protobuf::Message::GetTypeName() const'
../lib/libcaffe.so: undefined reference to `google::protobuf::Message::DebugString() const'
../lib/libcaffe.so: undefined reference to `google::protobuf::MessageLite::ParseFromString(std::string const&)'
../lib/libcaffe.so: undefined reference to `google::protobuf::internal::NameOfEnum(google::protobuf::EnumDescriptor const*, int)'
../lib/libcaffe.so: undefined reference to `google::protobuf::internal::ArenaStringPtr::AssignWithDefault(std::string const*, google::protobuf::internal::ArenaStringPtr)'
../lib/libcaffe.so: undefined reference to `google::protobuf::DescriptorPool::FindFileByName(std::string const&) const'
../lib/libcaffe.so: undefined reference to `google::protobuf::internal::WireFormatLite::ReadBytes(google::protobuf::io::CodedInputStream*, std::string*)'
../lib/libcaffe.so: undefined reference to `google::protobuf::MessageFactory::InternalRegisterGeneratedFile(char const*, void (*)(std::string const&))'
../lib/libcaffe.so: undefined reference to `leveldb::DB::Open(leveldb::Options const&, std::string const&, leveldb::DB**)'
../lib/libcaffe.so: undefined reference to `google::protobuf::internal::WireFormatLite::WriteBytesMaybeAliased(int, std::string const&, google::protobuf::io::CodedOutputStream*)'
../lib/libcaffe.so: undefined reference to `leveldb::Status::ToString() const'
../lib/libcaffe.so: undefined reference to `google::protobuf::internal::WireFormatLite::WriteString(int, std::string const&, google::protobuf::io::CodedOutputStream*)'
collect2: error: ld returned 1 exit status
tools/CMakeFiles/extract_features.dir/build.make:134: recipe for target 'tools/extract_features' failed
make[2]: *** [tools/extract_features] Error 1
CMakeFiles/Makefile2:433: recipe for target 'tools/CMakeFiles/extract_features.dir/all' failed
make[1]: *** [tools/CMakeFiles/extract_features.dir/all] Error 2
Makefile:129: recipe for target 'all' failed
make: *** [all] Error 2

roi_pooling_param {
    pooled_w: 7
    pooled_h: 7
    spatial_scale: 0.0625 # 1/16
  }

Looking at the layer, it seems to resize the output to [N, Channels, 7, 7] where N is the number of bounding boxes

I find an ambiguity in the way hstart, hend, wstart, wend are computed. When the pooling height and width are (3,3), the variable bin_size_h and bin_size_w are 2.667 (approximated to three places of decimal).
Then, there are two ways of computing hend. One is the current way, the alternative way being
hend = hstart + bin_size_h
When we compute that way for the dimension that I have mentioned above, (For ph = 1 and roi_start_h = 0. ),
hend = static_cast<int>(ceil(2.667)) + 2;
This gives the output as 5 whereas, the current method of computing in the caffe repository gives the output as 6. Since its a square window, the same applies for width as well.
Could someone please explain this discrepancy?

Hello, every Fast-rcnn player! I faced a problem while compiling the fast r-cnn, g++ compiler reports a function is not defined: caffe.cpp:(.text+0x20b): undefined reference to `caffe::Net::Net(std::string const&, caffe::Phase)'. Have anyone meet this problem too? And could you give me a hand? Thanks very much! The code is runing on Ubuntu14.04 x64 with g++4.8.4 compiling.

The detailed error information:

.build_release/tools/caffe.o: In function test()': caffe.cpp:(.text+0x20b): undefined reference tocaffe::Net::Net(std::string const&, caffe::Phase)'
.build_release/tools/caffe.o: In function train()': caffe.cpp:(.text+0x1395): undefined reference tocaffe::Caffe::singleton_'
.build_release/tools/caffe.o: In function time()': caffe.cpp:(.text+0x196c): undefined reference tocaffe::Net::Net(std::string const&, caffe::Phase)'
.build_release/tools/caffe.o: In function caffe::Caffe::Get()': caffe.cpp:(.text._ZN5caffe5Caffe3GetEv[_ZN5caffe5Caffe3GetEv]+0x7): undefined reference tocaffe::Caffe::singleton_'
.build_release/tools/caffe.o: In function caffe::Solver<float>* caffe::GetSolver<float>(caffe::SolverParameter const&)': caffe.cpp:(.text._ZN5caffe9GetSolverIfEEPNS_6SolverIT_EERKNS_15SolverParameterE[_ZN5caffe9GetSolverIfEEPNS_6SolverIT_EERKNS_15SolverParameterE]+0x4a): undefined reference tocaffe::Solver::Solver(caffe::SolverParameter const&)'
caffe.cpp:(.text._ZN5caffe9GetSolverIfEEPNS_6SolverIT_EERKNS_15SolverParameterE[_ZN5caffe9GetSolverIfEEPNS_6SolverIT_EERKNS_15SolverParameterE]+0x1b4): undefined reference to caffe::Solver<float>::Solver(caffe::SolverParameter const&)' caffe.cpp:(.text._ZN5caffe9GetSolverIfEEPNS_6SolverIT_EERKNS_15SolverParameterE[_ZN5caffe9GetSolverIfEEPNS_6SolverIT_EERKNS_15SolverParameterE]+0x2e4): undefined reference tocaffe::Solver::Solver(caffe::SolverParameter const&)'
caffe.cpp:(.text._ZN5caffe9GetSolverIfEEPNS_6SolverIT_EERKNS_15SolverParameterE[_ZN5caffe9GetSolverIfEEPNS_6SolverIT_EERKNS_15SolverParameterE]+0x3d4): undefined reference to caffe::Solver<float>::Solver(caffe::SolverParameter const&)' collect2: error: ld returned 1 exit status .build_release/tools/extract_features.o: In functioncaffe::Caffe::Get()':
extract_features.cpp:(.text._ZN5caffe5Caffe3GetEv[_ZN5caffe5Caffe3GetEv]+0x7): undefined reference to caffe::Caffe::singleton_' .build_release/tools/extract_features.o: In function int feature_extraction_pipeline(int, char**)':
extract_features.cpp:(.text._Z27feature_extraction_pipelineIfEiiPPc[Z27feature_extraction_pipelineIfEiiPPc]+0xa8): undefined reference tocaffe::Caffe::singleton_' extract_features.cpp:(.text._Z27feature_extraction_pipelineIfEiiPPc[_Z27feature_extraction_pipelineIfEiiPPc]+0x114): undefined reference to caffe::Net::Net(std::string const&, caffe::Phase)'
collect2: error: ld returned 1 exit status
make: *** [.build_release/tools/extract_features.bin] Error 1
make: *** Waiting for unfinished jobs....
make: *** [.build_release/tools/caffe.bin] Error 1
.build_release/tools/convert_imageset.o: In functionvoid caffe::shuffle<__gnu_cxx::__normal_iterator<std::pair<std::string, int>_, std::vector<std::pair<std::string, int>, std::allocator<std::pair<std::string, int> > > > >(__gnu_cxx::__normal_iterator<std::pair<std::string, int>_, std::vector<std::pair<std::string, int>, std::allocator<std::pair<std::string, int> > > >, __gnu_cxx::__normal_iterator<std::pair<std::string, int>*, std::vector<std::pair<std::string, int>, std::allocator<std::pair<std::string, int> > > >)': convert_imageset.cpp:(.text._ZN5caffe7shuffleIN9__gnu_cxx17__normal_iteratorIPSt4pairISsiESt6vectorIS4_SaIS4_EEEEEEvT_SA_[_ZN5caffe7shuffleIN9__gnu_cxx17__normal_iteratorIPSt4pairISsiESt6vectorIS4_SaIS4_EEEEEEvT_SA_]+0x14): undefined reference to caffe::Caffe::singleton'
collect2: error: ld returned 1 exit status
/usr/bin/ld: .build_release/examples/cpp_classification/classification.o: undefined reference to symbol '_ZN2cv6imreadERKNS_6StringEi'
//usr/local/lib/libopencv_imgcodecs.so.3.0: error adding symbols: DSO missing from command line
collect2: error: ld returned 1 exit status
make: *** [.build_release/examples/cpp_classification/classification.bin] Error 1
make: *** [.build_release/tools/convert_imageset.bin] Error 1

Attempting to build caffe-fast-rcnn with cuDNN v5 leads to these types of errors:

In file included from ./include/caffe/util/device_alternate.hpp:40:0,
                 from ./include/caffe/common.hpp:19,
                 from src/caffe/util/upgrade_proto.cpp:8:
./include/caffe/util/cudnn.hpp: In function ‘void caffe::cudnn::createPoolingDesc(cudnnPoolingStruct**, caffe::PoolingParameter_PoolMethod, cudnnPoolingMode_t*, int, int, int, int, int, int)’:
./include/caffe/util/cudnn.hpp:127:41: error: too few arguments to function ‘cudnnStatus_t cudnnSetPooling2dDescriptor(cudnnPoolingDescriptor_t, cudnnPoolingMode_t, cudnnNanPropagation_t, int, int, int, int, int, int)’
         pad_h, pad_w, stride_h, stride_w));
                                         ^
./include/caffe/util/cudnn.hpp:15:28: note: in definition of macro ‘CUDNN_CHECK’
     cudnnStatus_t status = condition; \

Any plans to update caffe-fast-rcnn to support cuDNN v5 and the newer Pascal GPUs?

Seems like the main caffe just recently added support for cuDNN v5 per BVLC/caffe#3969

Thanks!

When I compile this project, I got:

.build_release/tools/caffe.o: In function `train()':
caffe.cpp:(.text+0x1ab7): undefined reference to `caffe::P2PSync<float>::run(std::vector<int, std::allocator<int> > const&)'
collect2: error: ld returned 1 exit status
/usr/bin/ld: warning: libhdf5_hl.so.7, needed by /usr/local/lib/libcaffe.so, may conflict with libhdf5_hl.so.10
make: *** [.build_release/tools/caffe.bin] Error 1
make: *** Waiting for unfinished jobs....
CXX/LD -o .build_release/tools/train_net.bin

What's the problem