tf branch questions

[chenming-wu]

Hi Peng-Shuai,

I encountered two problems during using O-CNN tensorflow branch:

• there is a points_database = _tf_ocnn_module.points_database # todo: delete this operator in /tensorflow/libs/__init__.py, is that ok to delete this line? otherwise I cannot compile the lib.
• in /tensorflow/script/cls.py, it uses points_dataset in train_network function while the dataset is octree_dataset in test_network, is that correct or a typo?

Thanks!

[CRISZJ]

Hi Peng-Shuai,

I encountered two problems during using O-CNN tensorflow branch:

• there is a points_database = _tf_ocnn_module.points_database # todo: delete this operator in /tensorflow/libs/__init__.py, is that ok to delete this line? otherwise I cannot compile the lib.
• in /tensorflow/script/cls.py, it uses points_dataset in train_network function while the dataset is octree_dataset in test_network, is that correct or a typo?

Thanks!

hi,what's your problem, I also have some problems, it is as follows:

/home/user/anaconda3/envs/ocnn/lib/python3.6/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: libocnn.so: undefined symbol: _ZNK10PointsInfo12check_formatERSs

Did you have the same problem?

[chenming-wu]

Hi Peng-Shuai,
I encountered two problems during using O-CNN tensorflow branch:

• there is a points_database = _tf_ocnn_module.points_database # todo: delete this operator in /tensorflow/libs/__init__.py, is that ok to delete this line? otherwise I cannot compile the lib.
• in /tensorflow/script/cls.py, it uses points_dataset in train_network function while the dataset is octree_dataset in test_network, is that correct or a typo?

Thanks!

hi,what's your problem, I also have some problems, it is as follows:

/home/user/anaconda3/envs/ocnn/lib/python3.6/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: libocnn.so: undefined symbol: _ZNK10PointsInfo12check_formatERSs

Did you have the same problem?

Sorry, I don't have this problem. Maybe you could try to clean your built files and rebuild again.

[CRISZJ]

Hi Peng-Shuai,
I encountered two problems during using O-CNN tensorflow branch:

• there is a points_database = _tf_ocnn_module.points_database # todo: delete this operator in /tensorflow/libs/__init__.py, is that ok to delete this line? otherwise I cannot compile the lib.
• in /tensorflow/script/cls.py, it uses points_dataset in train_network function while the dataset is octree_dataset in test_network, is that correct or a typo?

Thanks!

hi,what's your problem, I also have some problems, it is as follows:
/home/user/anaconda3/envs/ocnn/lib/python3.6/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: libocnn.so: undefined symbol: _ZNK10PointsInfo12check_formatERSs
Did you have the same problem?

Sorry, I don't have this problem. Maybe you could try to clean your built files and rebuild again.

Thanks for your reply, I try to rebuild it , but is is not success. When you follow the author' readme, have you meeting this warning?

octree/octree/octree_nn.cu(329): warning: expression has no effect
octree/octree/octree_nn.cu(358): warning: expression has no effect
octree/octree/octree_nn.cu(383): warning: expression has no effect

[wang-ps]

@CRISZJ The warnings can be safely ignored.
Your problem on undefined symbol: _ZNK10PointsInfo12check_formatERSs is caused by the incompatible GCC version you used to complie the code.
Please EXACTLY follow the readme to build the code, if you still have problems, let me know.

[CRISZJ]

@CRISZJ The warnings can be safely ignored.
Your problem on undefined symbol: _ZNK10PointsInfo12check_formatERSs is caused by the incompatible GCC version you used to complie the code.
Please EXACTLY follow the readme to build the code, if you still have problems, let me know.
Thanks for your reply. The version of gcc in my computer is 5.4.0. Is that right? I also follow the reademe to build anthoer enviroments.

[wang-ps]

Do you install tensorflow via the following command: conda create -n tf-1.14.0 tensorflow-gpu==1.14.0

[CRISZJ]

Do you install tensorflow via the following command: conda create -n tf-1.14.0 tensorflow-gpu==1.14.0

Sure , I also use the virtual enviroments

[iDzh]

May I ask if anyone has run the code through?

[songyadong106]

Hello, I would like to ask you to follow your steps but can't get the label, can you help me to see it?The segmentation data of "02691156",thanks!

[songyadong106]

@chenming-wu @CRISZJ @wang-ps @iDzh
can you help me to see it?

[wang-ps]

@chenming-wu @songyadong106
Sorry for the delay. I have updated the readme and tested the tensorflow code just now.
Now you can follow the installation guide and shape classafication guide to have a try.