nvlabs / splatnet Goto Github PK
View Code? Open in Web Editor NEWSPLATNet: Sparse Lattice Networks for Point Cloud Processing (CVPR2018)
License: Other
SPLATNet: Sparse Lattice Networks for Point Cloud Processing (CVPR2018)
License: Other
I have reinstall the nvidia-docker with the following steps.
**sudo apt-get install -y nvidia-docker2
sudo docker pull suhangpro/caffe:bpcn
sudo docker run -it suhangpro/caffe:bpcn
pip install protobuf, scikit-image, bumpy
git clone https://github.com/NVlabs/splatnet
export PYTHONPATH=...
cd /workspace/splatnet/exp/shapenet3d
sh test_only.sh**
Then I got
test_only.sh: 6: test_only.sh: Bad substitution
WARNING: Logging before InitGoogleLogging() is written to STDERR
E0916 21:30:26.347004 133 common.cpp:114] Cannot create Cublas handle. Cublas won't be available.
E0916 21:30:26.347118 133 common.cpp:121] Cannot create Curand generator. Curand won't be available.
F0916 21:30:26.347223 133 common.cpp:152] Check failed: error == cudaSuccess (35 vs. 0) CUDA driver version is insufficient for CUDA runtime version
*** Check failure stack trace: ***
Aborted (core dumped)
Did I wrongly operate on anything?
hello, I follow the step in INSTALL.md to install dependency,. when I install nccl using command 'sudo make -j install ' it didn't work, and it shows ' make: *** No rule to make target 'install'. Stop.'. can you please tell me how to solve it. thanks very much
Hi, when I cd /splatnet/exp/facade3d, run SKIP_TRAIN=1 ./train_test.sh, it reports error as follows:
Traceback (most recent call last):
File "/home/ryxu/splatnet/exp/facade3d/../../splatnet/semseg3d/test.py", line 11, in
import splatnet.configs
ImportError: No module named 'splatnet'
I think maybe some wrong operation about 2.Include the project to your python path so imports can be found, I set my ./bashrc file as follows:
export PYCAFFE_ROOT=/home/ryxu/caffe/python
export PYTHONPATH=$PYCAFFE_ROOT:$PYTHONPATH
export PATH=$CAFFE_ROOT/build/tools:$PYCAFFE_ROOT:$PATH
##Include the project to your python path
export PYTHONPATH=</home/ryxu/splatnet/splatnet>:$PYTHONPATH
Could anyone can help me? thanks very much.
when I run facade test pre-trained model code, it occurs indexError as follows:
File "/home/yt/splatnet/exp/facade3d/../../splatnet/semseg3d/test.py", line 211, in
args.dataset_params, save_dir, save_prefix, args.cpu)
File "/home/yt/splatnet/exp/facade3d/../../splatnet/semseg3d/test.py", line 96, in semseg_test
data, xyz, norms = dataset_facade.points(dims=input_dims+',x_y_z,nx_ny_nz', **dataset_params)
File "/home/yt/splatnet/splatnet/dataset/dataset_facade.py", line 98, in points
return tuple([pcl_data[:, idx] * sc for (idx, sc) in zip(feat_idxs, feat_scales)])
File "/home/yt/splatnet/splatnet/dataset/dataset_facade.py", line 98, in
return tuple([pcl_data[:, idx] * sc for (idx, sc) in zip(feat_idxs, feat_scales)])
IndexError: index 6 is out of bounds for axis 1 with size 6
I don,t know whether my installation is well installed
Thank you for the wonderful work.But I have run into an error I could not find solution.
My environment is ubuntu16.04,cuda9.2,cudnn7.6.4,pytorch1.1.0.
I can train the dataset in ShapeNet such as ‘bag’,'cap','earphone' and etc.
Their item numbes are under 500.
But when I tried to train the 'airplane','chair','table'an etc. I find the error below:
I cannot find where is the codes to generate this file.
I checked the memory and the video memory,and they didn't reach the upper limit.
Thank you.
Could you tell me when could we expect the release of joint 2D-3D experiments part of your research ?
I‘m interested in your great work, and I want to train with another dataset with 12 classes, I am confused at what I should change about training with differentnumber except for classed number and cmap? Can you please help me ?? thanks very much, waiting for your reply
Hi,
May I know when will the Joint 2D-3D experiments become available to try.
Thanks!
William
May I ask some issues about "module 'caffe.proto.caffe_pb2' has no attribute 'PermutohedralParameter'"?
Is there a problem with my Caffe version?
Hi, when I try to run the command: SKIP_TRAIN=1 ./exp/facade3d/train_test.sh, it has problem as title:
Traceback (most recent call last):
File "/home/ryxu/splatnet/exp/facade3d/../../splatnet/semseg3d/test.py", line 211, in
args.dataset_params, save_dir, save_prefix, args.cpu)
File "/home/ryxu/splatnet/exp/facade3d/../../splatnet/semseg3d/test.py", line 96, in semseg_test
data, xyz, norms = dataset_facade.points(dims=input_dims+',x_y_z,nx_ny_nz', **dataset_params)
File "/home/ryxu/splatnet/splatnet/dataset/dataset_facade.py", line 90, in points
pcl_data = np.loadtxt(pcl_test_path, skiprows=15)
File "/home/ryxu/miniconda3/envs/caffe/lib/python3.5/site-packages/numpy/lib/npyio.py", line 880, in loadtxt
next(fh)
File "/home/ryxu/miniconda3/envs/caffe/lib/python3.5/codecs.py", line 321, in decode
(result, consumed) = self._buffer_decode(data, self.errors, final)
UnicodeDecodeError: 'utf-8' codec can't decode byte 0xbc in position 334: invalid start byte
It maybe the decode problem, how to solve this problem in this splatnet project? Thanks.
Hello ,may I konw when the part of Joint 2D-3D is available?Thanks~
what if I don't have attribute-height, and i don't want to use height value for training, what should I do? any way can deal with it?
Hi, suhangpro
Thanks for your great contribution and efforts.
When I attempted to run your splatnet code with the docker image you pushed on my pc with an NVIDIA 2080Ti, it turned out that the cuda version is incompatible with my hardware.
So I managed to create a docker image that supports NVIDIA 2080Ti (Turing Arcitechture), with cuda10.0+cudnn7+caffe+bilateralNN+conda+python3.5+ubuntu16.04. Actually the only difference with your image is the CUDA support update.
Consider to add this image
https://hub.docker.com/repository/docker/zyzwhdx/splatnet/general
Thanks!
Shouldn't accuracy= TPs / (TPs+FPs)? What computed in below two lines seem to be recall.
splatnet/splatnet/semseg3d/eval_seg.py
Line 38 in f821913
Line 149 in f821913
Would there still be a plan to release joint 2D-3D experiments? If so, may I ask what is the estimated release time? Sorry for the duplicated issue.
I am new to docker environment. Basically, I am able to enter the docker container and use caffe. However, when I run the test script, it complains that the cuda driver version is not sufficient to cuda runtime. What is the required cuda driver? And how can I update it in docker? Thanks.
For semantic segmentation, ruemonge428 dataset is post-processed, height value is added to each point, I want to know the definition of height to each point. thanks very much!
My server GPU has a problem. So I can not use GPU for training and test for now,
How I can disable GPU for training and test?
Any suggestions? Thanks!
@foertter @dumerrill @tmbdev thanks for open sourcing the wonderfull work , i had few queries
Q1 have you trained the architecture on the available other dataset like semanttic Kitti and 3D dataset
Q2 If not trained can we follow the same training pipeline , if trained can you please share the pre-trained model
Q3 can we use the currently pre-trained model to test on custom dataset which less number of point cloud density
Thanks in advance
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.