There is a question when I run the demo.

[zhoucong666666]

when I make in the lib,there is an error:/usr/local/lib/python2.7/dist-packages/tensorflow/include/tensorflow/core/platform/default/mutex.h:25:22: fatal error: nsync_cv.h: No such file or directory. could you help me? thank you!

[JGuillaumin]

Same issue this morning ...

[Kongsea]

You can find this file and replace it with the absolute path.

[JGuillaumin]

Yes. You have to edit : /usr/local/lib/python2.7/dist-packages/tensorflow/include/tensorflow/core/platform/default/mutex.h (path mentioned in your error message).

By replacing :

#include "nsync_cv.h"
#include "nsync_mu.h"

By :

#include "external/nsync/public/nsync_cv.h"
#include "external/nsync/public/nsync_mu.h"

Now, make command works !

Or you can edit make.sh. This error is due to the new placement of include files for external libraries in TensorFlow ! (from 1.1).

[JGuillaumin]

I prefer to modify make.sh instead of modifying TensorFlow source code :

TF_INC=$(python -c 'import tensorflow as tf; print(tf.sysconfig.get_include())')
NSYNC_INC=$TF_INC"/external/nsync/public"

CUDA_PATH=/usr/local/cuda/
CXXFLAGS=''

if [[ "$OSTYPE" =~ ^darwin ]]; then
	CXXFLAGS+='-undefined dynamic_lookup'
fi

cd roi_pooling_layer

if [ -d "$CUDA_PATH" ]; then
	nvcc -std=c++11 -c -o roi_pooling_op.cu.o roi_pooling_op_gpu.cu.cc \
		-I $TF_INC -I $NSYNC_INC -D GOOGLE_CUDA=1 -x cu -Xcompiler -fPIC $CXXFLAGS \
		-arch=sm_37

	g++ -std=c++11 -shared -o roi_pooling.so roi_pooling_op.cc \
		roi_pooling_op.cu.o -I $TF_INC  -I $NSYNC_INC -D GOOGLE_CUDA=1 -fPIC $CXXFLAGS \
		-lcudart -L $CUDA_PATH/lib64
else
	g++ -std=c++11 -shared -o roi_pooling.so roi_pooling_op.cc \
		-I $TF_INC -I $NSYNC_INC -fPIC $CXXFLAGS
fi

cd ..

I added NSYNC_INC variable, that correspond to the directory where nsync_cv.h is !

It works !

[JGuillaumin]

Pull request : https://github.com/smallcorgi/Faster-RCNN_TF/pull/246/commits/218d17a15c2d55e093fd8f41c545744c8b7faebc

[JGuillaumin]

In addition, you can encounter some Undefined symbol error, while the compilation was done correctly. Same issue, it's due to new compilation rules since Tensorflow 1.x.y !

Correct make file ( for tensorflow 1.4.0-gpu) :

TF_INC=$(python -c 'import tensorflow as tf; print(tf.sysconfig.get_include())')
TF_LIB=$(python -c 'import tensorflow as tf; print(tf.sysconfig.get_lib())')
NSYNC_INC=$TF_INC"/external/nsync/public"

CUDA_PATH=/usr/local/cuda/
CXXFLAGS='-D_GLIBCXX_USE_CXX11_ABI=0'

if [[ "$OSTYPE" =~ ^darwin ]]; then
	CXXFLAGS+='-undefined dynamic_lookup'
fi

cd roi_pooling_layer

if [ -d "$CUDA_PATH" ]; then
	nvcc -std=c++11 -c -o roi_pooling_op.cu.o roi_pooling_op_gpu.cu.cc \
		-I $TF_INC -I $NSYNC_INC -D GOOGLE_CUDA=1 -x cu -Xcompiler -fPIC $CXXFLAGS \
		-arch=sm_37 -L $TF_LIB -ltensorflow_framework

	g++ -std=c++11 -shared -o roi_pooling.so roi_pooling_op.cc \
		roi_pooling_op.cu.o -I $TF_INC  -I $NSYNC_INC -D GOOGLE_CUDA=1 -fPIC $CXXFLAGS \
		-lcudart -L $CUDA_PATH/lib64 -L $TF_LIB -ltensorflow_framework
else
	g++ -std=c++11 -shared -o roi_pooling.so roi_pooling_op.cc \
		-I $TF_INC -I $NSYNC_INC -fPIC $CXXFLAGS -L $TF_LIB -ltensorflow_framework
fi

cd ..

I changed CXXFLAGS='-D_GLIBCXX_USE_CXX11_ABI=0', and I added a new compilation flag -L $TF_LIB -ltensorflow_framework with TF_LIB=$(python -c 'import tensorflow as tf; print(tf.sysconfig.get_lib())') !

Now demo.py and training script can work ! (about 0.226s/iter on a GTX 1070 for training).

[zhoucong666666]

I had changed make.sh,but there is another error: roi_pooling_op.cc:21:61: fatal error: third_party/eigen3/unsupported/Eigen/CXX11/Tensor: No such file or directory

and I try some methods: CXXFLAGS+='-undefined dynamic_lookup' to CXXFLAGS='-D_MWAITXINTRIN_H_INCLUDED'

but,it still dose not work.

[JGuillaumin]

Which version of TensorFlow are you using ? On linux ? Built from binaries ? or from source code ?

[zhoucong666666]

tensorflow1.4.0 on linux

[JGuillaumin]

hummm .. it seems curious.

1/ Did you activate your python env where TensorFlow is installed ?? (like source activate tensorflow)

2 / Is there third_party/eigen3/unsupported/Eigen/CXX11/Tensor starting from this path TF_INC=$(python -c 'import tensorflow as tf; print(tf.sysconfig.get_include())') ??

3/ Can you try to reinstall Tensorflow :

pip install --ignore-installed --upgrade https://storage.googleapis.com/tensorflow/linux/cpu/tensorflow-1.4.0-cp35-cp35m-linux_x86_64.whl

(here for CPU and Python 3.5, see on TensorFlow Install page, to find the URL from different version of Python and CPU vs GPU.

[ambr89]

I've downgraded tensorflow to 1.3 and I resolve this problem

[AliceLcz]

I have the same problem as @zhoucong666666 ,when I do make in lib, error like this: ~/Faster-RCNN_TF/lib$ make python setup.py build_ext --inplace running build_ext skipping 'utils/bbox.c' Cython extension (up-to-date) skipping 'utils/nms.c' Cython extension (up-to-date) skipping 'nms/cpu_nms.c' Cython extension (up-to-date) skipping 'nms/gpu_nms.cpp' Cython extension (up-to-date) rm -rf build bash make.sh Traceback (most recent call last): File "", line 1, in ImportError: No module named tensorflow roi_pooling_op.cc:21:61: fatal error: third_party/eigen3/unsupported/Eigen/CXX11/Tensor: No such file or directory compilation terminated. But I do have the file 'Tensor' in home/tensorflow/third_party/eigen3/unsupported/Eigen/CXX11/Tensor, I don't know why, may I get your some help? @JGuillaumin Thanks very much!!!

[roachsinai]

@JGuillaumin ths.

[roachsinai]

But @JGuillaumin, could you give me some advice as may tensorflow version is 1.6.0.

Got this error when I run demo.py:

Traceback (most recent call last):
  File "./tools/demo.py", line 11, in <module>
    from networks.factory import get_network
  File "/home/roach/code/Git/Faster-RCNN_TF/tools/../lib/networks/__init__.py", line 8, in <module>
    from .VGGnet_train import VGGnet_train
  File "/home/roach/code/Git/Faster-RCNN_TF/tools/../lib/networks/VGGnet_train.py", line 2, in <module>
    from networks.network import Network
  File "/home/roach/code/Git/Faster-RCNN_TF/tools/../lib/networks/network.py", line 3, in <module>
    import roi_pooling_layer.roi_pooling_op as roi_pool_op
  File "/home/roach/code/Git/Faster-RCNN_TF/tools/../lib/roi_pooling_layer/roi_pooling_op.py", line 5, in <module>
    _roi_pooling_module = tf.load_op_library(filename)
  File "/usr/lib/python3.6/site-packages/tensorflow/python/framework/load_library.py", line 58, in load_op_library
    lib_handle = py_tf.TF_LoadLibrary(library_filename, status)
  File "/usr/lib/python3.6/site-packages/tensorflow/python/framework/errors_impl.py", line 516, in __exit__
    c_api.TF_GetCode(self.status.status))
tensorflow.python.framework.errors_impl.NotFoundError: /home/roach/code/Git/Faster-RCNN_TF/tools/../lib/roi_pooling_layer/roi_pooling.so: undefined symbol: _Z22ROIPoolBackwardLaucherPKffiiiiiiiS0_PfPKiRKN5Eigen9GpuDeviceE

Or, I have to downgrade my tensorflw, :-(

[fanw52]

Hello,I use the Makefile that you present,and then I can't find -ltensorflow-framework,how can you do it? thx? @JGuillaumin

[cjb752815359]

But @JGuillaumin, could you give me some advice as may tensorflow version is 1.6.0.

Got this error when I run demo.py:

Traceback (most recent call last):
  File "./tools/demo.py", line 11, in <module>
    from networks.factory import get_network
  File "/home/roach/code/Git/Faster-RCNN_TF/tools/../lib/networks/__init__.py", line 8, in <module>
    from .VGGnet_train import VGGnet_train
  File "/home/roach/code/Git/Faster-RCNN_TF/tools/../lib/networks/VGGnet_train.py", line 2, in <module>
    from networks.network import Network
  File "/home/roach/code/Git/Faster-RCNN_TF/tools/../lib/networks/network.py", line 3, in <module>
    import roi_pooling_layer.roi_pooling_op as roi_pool_op
  File "/home/roach/code/Git/Faster-RCNN_TF/tools/../lib/roi_pooling_layer/roi_pooling_op.py", line 5, in <module>
    _roi_pooling_module = tf.load_op_library(filename)
  File "/usr/lib/python3.6/site-packages/tensorflow/python/framework/load_library.py", line 58, in load_op_library
    lib_handle = py_tf.TF_LoadLibrary(library_filename, status)
  File "/usr/lib/python3.6/site-packages/tensorflow/python/framework/errors_impl.py", line 516, in __exit__
    c_api.TF_GetCode(self.status.status))
tensorflow.python.framework.errors_impl.NotFoundError: /home/roach/code/Git/Faster-RCNN_TF/tools/../lib/roi_pooling_layer/roi_pooling.so: undefined symbol: _Z22ROIPoolBackwardLaucherPKffiiiiiiiS0_PfPKiRKN5Eigen9GpuDeviceE

Or, I have to downgrade my tensorflw, :-(

Do you solve this problem? I have the same problem. Do you run demo on gpu?