tensorflow.python.framework.errors_impl.NotFoundError

[jousefm]

Did anyone encounter this issue so far when running demo.py?

File "/root/miniconda3/envs/../lib/python3.7/importlib/__init__.py", line 127, in import_module
    return _bootstrap._gcd_import(name[level:], package, level)
  File "<frozen importlib._bootstrap>", line 1006, in _gcd_import
  File "<frozen importlib._bootstrap>", line 983, in _find_and_load
  File "<frozen importlib._bootstrap>", line 967, in _find_and_load_unlocked
  File "<frozen importlib._bootstrap>", line 677, in _load_unlocked
  File "<frozen importlib._bootstrap_external>", line 728, in exec_module
  File "<frozen importlib._bootstrap>", line 219, in _call_with_frames_removed
  File "/workspace/../pcn/models/pcn_cd.py", line 26, in <module>
    from tf_util import *
  File "/workspace/../pcn/tf_util.py", line 26, in <module>
    from pc_distance import tf_nndistance, tf_approxmatch
  File "/workspace/../pcn/pc_distance/tf_nndistance.py", line 5, in <module>
    nn_distance_module=tf.load_op_library(os.path.join(BASE_DIR, 'tf_nndistance_so.so'))
  File "/root/miniconda3/envs/../lib/python3.7/site-packages/tensorflow/python/framework/load_library.py", line 61, in load_op_library
    lib_handle = py_tf.TF_LoadLibrary(library_filename)
tensorflow.python.framework.errors_impl.NotFoundError: /workspace/../pcn/pc_distance/tf_nndistance_so.so: undefined symbol: _ZN10tensorflow12OpDefBuilder5InputESs

[Grace1206]

I also met this problem, have you already solved it?

[MarioCavero]

In my case when using Ubuntu 18.04, with updated CUDA, cudnn and adding the proper funciton operator in C++ (due to incompatibility)

    @ops.RegisterShape('ApproxMatch')
AttributeError: module 'tensorflow.python.framework.ops' has no attribute 'RegisterShape'

I got undefined symbol:_ZN10tensorflow8OpKernel11TraceStringB5cxx11EPNS_15OpKernelContextEb . This specific error means that there are CUDA incompatibilities, so the versionings are wrong, so make sure you are using the exact requirements and system this repo uses!

This code is built using Tensorflow 1.12 with CUDA 9.0 and tested on Ubuntu 16.04 with Python 3.5.

[ybhhdt]

In my case when using Ubuntu 18.04, with updated CUDA, cudnn and adding the proper funciton operator in C++ (due to incompatibility)

    @ops.RegisterShape('ApproxMatch')
AttributeError: module 'tensorflow.python.framework.ops' has no attribute 'RegisterShape'

I got undefined symbol:_ZN10tensorflow8OpKernel11TraceStringB5cxx11EPNS_15OpKernelContextEb . This specific error means that there are CUDA incompatibilities, so the versionings are wrong, so make sure you are using the exact requirements and system this repo uses!

This code is built using Tensorflow 1.12 with CUDA 9.0 and tested on Ubuntu 16.04 with Python 3.5.

Maybe,this is the better way to solve this problem. Tensorflow is not a good choice for researchers.