compile error on running tf_sampling_compile.sh

[amiltonwong]

Hi, authors,

I got error when I compile the fps code by commands:
sh tf_sampling_compile.sh

Here is the error log:

tf_sampling.cpp:6:42: fatal error: tensorflow/core/framework/op.h: No such file or directory
compilation terminated.

My system is ubuntu 16.04, with tf1.4 (python 2.7.14)

Any suggestion to fix it? THX!

[mf27]

Please check if you updated the compile script according to:
https://github.com/charlesq34/pointnet2#compile-customized-tf-operators

[WenshuangSong]

My system is ubuntu 16.04, with tf1.6, cuda9.1 (python 2.7.14)
And the same issues,
should I install cuda8.0 or tf 1.2 ?
ths!

[romanskie]

Hey Guys,
any solution on that?
Got the same issue on Ubuntu 16.04. I already tested with TF 1.4 (Cuda 8) and TF 1.8 (Cuda 8 / Cuda 9) on Python 2.7 plus Python 3.5 as well, but it did not work out. Maybe there is some sort oft work around here to make it work? I already tried the compile script modifications from the readme.

BG, Roman

[romanskie]

Hey, I was able to solve my problem. Somehow my local TF path was not the same as used in the .sh script, which leads to the "missing" c++ header files. Maybe I messed up the installation of TF with pip.

Changed the -I /usr/local/lib/python2.7/dist-packages/tensorflow/include in the script to /home/$MYUSERNAME/.local/lib/python2.7/site-packages/tensorflow/include and it worked.

[chenjunboBUPT]

Maybe you should check your path for tensorflow.

[leejiajun]

you should check your path for tensorflow.
dist-packages or site-packages
I made a mistake in spelling, GOD.

[iris0329]

as leejiajun said,
in tf sampling compile.sh file, change your tensorflow path. for example, I am using virtual env, so I change /usr/local/lib/python2.7/dist-packages/tensorflow/include to /....../py2/venv_python2.7/lib/python2.7/site-packages/tensorflow/include(my tensorflow path)
such changes will work

[MPashaei]

Hey Guys

I am using Google colab to run the code. I installed tensorflow-gpu==2.0.0-beta1. I just changed the tensorflow path in .sh files and .cpp files.
I am getting the error below.
tf_sampling.cpp:6:10: fatal error: /usr/local/lib/python3.6/dist-packages/tensorflow/core/framework/op.h: No such file or directory
#include "/usr/local/lib/python3.6/dist-packages/tensorflow/core/framework/op.h"
^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
compilation terminated.
I am wondering if I should change the tensorflow path in .cpp files as well.

When I use the same path in .cpp files, I still get the same error

tf_sampling.cpp:6:10: fatal error: tensorflow/core/framework/op.h: No such file or directory
#include "tensorflow/core/framework/op.h"

[mhs1212010]

hello
python singleton.py conll05_final_wsj
Traceback (most recent call last):
File "singleton.py", line 16, in
from srl_model import SRLModel
File "/home/moh/Desktop/unisrl-master/srl_model.py", line 15, in
from model_utils import *
File "/home/moh/Desktop/unisrl-master/model_utils.py", line 6, in
import srl_ops
File "/home/moh/Desktop/unisrl-master/srl_ops.py", line 4, in
srl_op_library = tf.load_op_library("./srl_kernels.so")
File "/usr/local/lib/python2.7/dist-packages/tensorflow/python/framework/load_library.py", line 58, in load_op_library
lib_handle = py_tf.TF_LoadLibrary(library_filename, status)
File "/usr/local/lib/python2.7/dist-packages/tensorflow/python/framework/errors_impl.py", line 516, in exit
c_api.TF_GetCode(self.status.status))
tensorflow.python.framework.errors_impl.NotFoundError: ./srl_kernels.so: undefined symbol: _ZN10tensorflow8internal21CheckOpMessageBuilder9NewStringB5cxx11Ev

My system is ubuntu 16.04, with tf1.8, cuda9.1 (python 2.7.14)
And the same issues,

[prajaktasurvase6]

Hey Guys

I am using Google colab to run the code. I installed tensorflow-gpu==2.0.0-beta1. I just changed the tensorflow path in .sh files and .cpp files.
I am getting the error below.
tf_sampling.cpp:6:10: fatal error: /usr/local/lib/python3.6/dist-packages/tensorflow/core/framework/op.h: No such file or directory
#include "/usr/local/lib/python3.6/dist-packages/tensorflow/core/framework/op.h"
^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
compilation terminated.
I am wondering if I should change the tensorflow path in .cpp files as well.

When I use the same path in .cpp files, I still get the same error

tf_sampling.cpp:6:10: fatal error: tensorflow/core/framework/op.h: No such file or directory
#include "tensorflow/core/framework/op.h"

Did you get any solution?

[AllenPu]

�I have a quick solution, under your virtual environment and run : python
then import tensorflow as tf; tf.path;
find your tensorflow path and change the first -I path from “/usr/local/lib/python2.7/dist-packages/tensorflow/include” to the path returned before.

[AllenPu]

�I have a quick solution, under your virtual environment and run : python
then import tensorflow as tf; tf.path;
find your tensorflow path and change the first -I path from “/usr/local/lib/python2.7/dist-packages/tensorflow/include” to the path returned before.

also useful for the other code based on pointnet: frsutum pointnet and so on.

[aradhyamathur]

@prajaktasurvase6 it seems that when I checked in docker which I was using as setup for this , the include folder was in tensorflow_core, so in the compilation script I had to change the path to tensorflow_core/include , it did resolve the issue you pointed out but led to another issue pointed out by @mhs1212010 here