dlinzhao / jsnet Goto Github PK
View Code? Open in Web Editor NEWJSNet: Joint Instance and Semantic Segmentation of 3D Point Clouds, AAAI2020
License: MIT License
JSNet: Joint Instance and Semantic Segmentation of 3D Point Clouds, AAAI2020
License: MIT License
hello,I successfully trained my data set. My dataset has only 2 categories. I tried to modify the code and changed 13 to 2. The errors reported during the test are as follows. What's the problem?
f = h5py.File(filename)
Traceback (most recent call last):
File "test.py", line 281, in
test()
File "test.py", line 159, in test
cur_pred_sem_softmax = np.zeros([cur_sem.shape[0], cur_sem.shape[1], NUM_CLASSES])
IndexError: tuple index out of range
I look forward to your reply very much, and I will be very grateful!!!
Thank you for your sharing of the codes.But when I run train.py,I got problem that the program was running all the time.I found the program maybe stuck in this sentence:
for batch_idx in range(num_batches):
print('here5--------------')
current_data, current_sem, current_label = dataset.get_batch(False)
Have you any idea for why could this be happening?Look forward to your reply.
Thanks for your sharing. I have trained the code on S3DIS very well, but I don't know how to train it on ShapeNet as the paper. Cou you give me some advice?
Why relative paths cannot be used?
Where to download datasets?
and
How to run it on macbook air m1, do I need to rent a server?
The following error will occur at the beginning of the program:
/root/autodl-tmp/JSNet/data/indoor3d_ins_seg_hdf5/Area_1_conferenceRoom_1.h5: (80, 4096, 9), (80, 4096), (80, 4096)
Traceback (most recent call last):
File "utils/s3dis_utils/s3dis_gen_h5.py", line 56, in
data_dtype, label_dtype)
File "/root/autodl-tmp/JSNet/utils/data_prep_util.py", line 111, in save_h5ins
h5_fout = h5py.File(h5_filename)
File "/root/miniconda3/envs/tf1/lib/python3.6/site-packages/h5py/_hl/files.py", line 427, in init
swmr=swmr)
File "/root/miniconda3/envs/tf1/lib/python3.6/site-packages/h5py/_hl/files.py", line 190, in make_fid
fid = h5f.open(name, flags, fapl=fapl)
File "h5py/_objects.pyx", line 54, in h5py._objects.with_phil.wrapper
File "h5py/_objects.pyx", line 55, in h5py._objects.with_phil.wrapper
File "h5py/h5f.pyx", line 96, in h5py.h5f.open
OSError: Unable to open file (unable to open file: name = '/root/autodl-tmp/JSNet/data/indoor3d_ins_seg_hdf5/Area_1_conferenceRoom_1.h5', errno = 2, error message = 'No such file or directory', flags = 0, o_flags = 0)
Error performance:
SOLUTION
After trying, I found that if I change h5py.file in the function save_h5ins to mode='a', it will run successfully
data_prep_util.py, line 111
h5_fout = h5py.File(h5_filename, mode='a')
END
Hoping that other beginners like me will not be troubled by this error.
When I use my own dataset,it throws error as following,i am sure my dataset path is correct,but it still .Could you please help me?
E:\Anaconda3\envs\lpb\python.exe D:/ASIS-master/gen_h5.py
D:\ASIS-master\data\indoor3d_ins_seg_hdf5\Area_1_hallway_1.h5: (3, 4096, 9), (3, 4096), (3, 4096)
D:\ASIS-master\utils\data_prep_util.py:95: H5pyDeprecationWarning: The default file mode will change to 'r' (read-only) in h5py 3.0. To suppress this warning, pass the mode you need to h5py.File(), or set the global default h5.get_config().default_file_mode, or set the environment variable H5PY_DEFAULT_READONLY=1. Available modes are: 'r', 'r+', 'w', 'w-'/'x', 'a'. See the docs for details.
h5_fout = h5py.File(h5_filename)
Traceback (most recent call last):
File "D:/ASIS-master/gen_h5.py", line 37, in
random_sample=False, sample_num=None)
File "D:\ASIS-master\indoor3d_util.py", line 281, in room2blocks_wrapper_normalized
random_sample, sample_num, sample_aug)
File "D:\ASIS-master\indoor3d_util.py", line 257, in room2blocks_plus_normalized
random_sample, sample_num, sample_aug)
File "D:\ASIS-master\indoor3d_util.py", line 215, in room2blocks
return np.concatenate(block_data_list, 0),
File "<array_function internals>", line 6, in concatenate
ValueError: need at least one array to concatenate
Process finished with exit code 1
Hello, may I ask if the instance labels are manually assigned? It seems we can obtain the instances label in those '.npy' files after 'collect_indoor3d_data.py'. I could not clearly understand how it gives us the instance labels.
Hello,Your work is excellent.I found that the coordinate system in the visual file generated by the test is different from that in the data set. I want the two coordinate systems to be the same. How can I operate? Your answer is very important to me. I look forward to your reply.It's important to me. Thank you in advance.
hello,
The following error occurred while making my own dataset:
Traceback (most recent call last):
File "utils/s3dis_utils/s3dis_gen_h5.py", line 36, in
random_sample=False, sample_num=None)
File "/shiyanshi2/user/sunyijun/code/JSNet-master-kiwifruit/utils/indoor3d_util.py", line 372, in room2blocks_wrapper_normalized
random_sample, sample_num, sample_aug)
File "/shiyanshi2/user/sunyijun/code/JSNet-master-kiwifruit/utils/indoor3d_util.py", line 343, in room2blocks_plus_normalized
random_sample, sample_num, sample_aug)
File "/shiyanshi2/user/sunyijun/code/JSNet-master-kiwifruit/utils/indoor3d_util.py", line 296, in room2blocks
return np.concatenate(block_data_list, 0),
File "<array_function internals>", line 6, in concatenate
ValueError: need at least one array to concatenate
Can you help me see what's wrong?
Hello Lin,
When I train this model,The error is reported as follows:
tensorflow.python.framework.errors_impl.NotFoundError: libtensorflow_framework.so: cannot open shared object file: No such file or directory
I've tried a lot of things.And it still hasn't been resolved.
pls,i need ur help
Hello People, I'm wondering how the authors of the paper tested their network on the ShapeNet dataset? Any ideas?
@dlinzhao When I run train.py, I get a error "tf_sampling_so.so not found". But I see the file in /tf_ops/sampling/tf_sampling_so.so. Why does train.py cannot find the file?
Do you know how to fix this?
in the paper, you said the sem would get through 1D Convolution with non-linear operation to the ins, and as it did in the code. However, the output of the net work has size of Bn13 and the ins have the size of bn5. i wonder Why is the output the opposite between ins and sem. i always thought the sem means the 13.
Could you tell me what your version of TensorFlow is?
Hi, thanks for your code. I tested the JSNet with my own data created from a Dynamic Vision Sensor. I get some weird test results, as you can see from the image. The Bounding Boxes are the instances, and the colors are the different semantic classes. Is it possible, like for instance 2, that points from different semantic classes are clustered to one instance? Moreover, I don’t understand why instance 1 has some points from instance 4 even if there is a huge space between them. Do you know any reason why this could happen? Did I miss something while clustering?
Thanks for your help!
About the timeliness,how long does it take to process a point cloud ?
Hello Lin,
Do I need to compile the .sh scripts under tf_ops first? Kindly advise on how to compile the tf operators.
Also did you test it in a specific cuda/cudnn version?
Hello, I want to ask how to separate each instance. For example, I just want to visualize each chair. What should I do?
Traceback (most recent call last):
File "train.py", line 226, in
train()
File "train.py", line 183, in train
train_one_epoch(sess, ops, train_writer, dataset, epoch)
File "train.py", line 222, in train_one_epoch
logger.info('mean loss: %f' % (loss_sum / float(num_batches)))
ZeroDivisionError: float division by zero
Exception ignored in: <bound method S3DISDataset.del of <s3dis_utils.dataset_s3dis.S3DISDataset object at 0x7f6acd354ac8>>
Traceback (most recent call last):
File "/shiyanshi2/user/sunyijun/code/JSNet-master-kiwifruit/utils/s3dis_utils/dataset_s3dis.py", line 151, in del
while not self.data_sample_queue.empty() and not self.data_queue.empty():
File "", line 2, in empty
File "/shiyanshi2/software/anaconda3/envs/JSNet/lib/python3.5/multiprocessing/managers.py", line 713, in _callmethod
self._connect()
File "/shiyanshi2/software/anaconda3/envs/JSNet/lib/python3.5/multiprocessing/managers.py", line 700, in _connect
conn = self._Client(self._token.address, authkey=self._authkey)
File "/shiyanshi2/software/anaconda3/envs/JSNet/lib/python3.5/multiprocessing/connection.py", line 487, in Client
c = SocketClient(address)
File "/shiyanshi2/software/anaconda3/envs/JSNet/lib/python3.5/multiprocessing/connection.py", line 614, in SocketClient
s.connect(address)
FileNotFoundError: [Errno 2] No such file or directory
What's wrong? We look forward to your reply
As title. Thank you !
Hi, thank you for your contribution to the open source cause.I would like to ask the prediction results file generated nine dimensional vector TXT file, what is their meaning?How do you make it as visual as in the paper?
It seems stop here all the time.Could you please help me ?
2022-05-28 13:08:41.829163: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1423] Adding visible gpu devices: 0
2022-05-28 13:08:42.016250: I tensorflow/core/common_runtime/gpu/gpu_device.cc:911] Device interconnect StreamExecutor with strength 1 edge matrix:
2022-05-28 13:08:42.016278: I tensorflow/core/common_runtime/gpu/gpu_device.cc:917] 0
2022-05-28 13:08:42.016284: I tensorflow/core/common_runtime/gpu/gpu_device.cc:930] 0: N
2022-05-28 13:08:42.016371: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1041] Created TensorFlow device (/job:localhost/replica:0/task:0/device:GPU:0 with 10975 MB memory) -> physical GPU (device: 0, name: NVIDIA TITAN X (Pascal), pci bus id: 0000:01:00.0, compute capability: 6.1)
A declarative, efficient, and flexible JavaScript library for building user interfaces.
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
An Open Source Machine Learning Framework for Everyone
The Web framework for perfectionists with deadlines.
A PHP framework for web artisans
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
Some thing interesting about web. New door for the world.
A server is a program made to process requests and deliver data to clients.
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Some thing interesting about visualization, use data art
Some thing interesting about game, make everyone happy.
We are working to build community through open source technology. NB: members must have two-factor auth.
Open source projects and samples from Microsoft.
Google ❤️ Open Source for everyone.
Alibaba Open Source for everyone
Data-Driven Documents codes.
China tencent open source team.