hszhao / pointweb Goto Github PK

View Code? Open in Web Editor NEW

199.0 199.0 18.0 336 KB

PointWeb: Enhancing Local Neighborhood Features for Point Cloud Processing, CVPR2019.

License: MIT License

Python 71.60% C++ 11.12% Cuda 16.72% Shell 0.56%

pointweb's People

Contributors

Stargazers

Watchers

Forkers

nnu-gisa hyzcn peterzhousz shnhrtkyk ahandsomeperson lfs119 erictuanle tangbohu haniitani sjingwen manhlab hetao255 ziyan-wyq ricemcm liuxinren456852 whuhxb geoeo roarou

pointweb's Issues

Config

Hi, Thank you for your great work!
I want to ask you how to create "config/s3dis/s3dis_pointweb.yaml", could you give me an example?
Thank you very much!

How do you do 3d visualization？

Confusing lists in the provided data

There is no 'val5_full.txt' required by 's3dis_pointweb.yaml' in the provided data and what are the 'train5.txt' and 'val6.txt' used for? Besides, how can I test the 6-folds cross validation?

how to install setup.py file for windows user?

when i run python setup.py install command, i meet this error:
error: command 'E:\vc2017\VC\Tools\MSVC\14.16.27023\bin\HostX86\x64\link.exe' failed with exit status 1120

Could you provide the model of ModelNet40?

I am interested in this paper.I hope you can provide the model of ModelNer40.Thank you very much.

File "/home/user/anaconda3/lib/python3.7/site-packages/torch/utils/cpp_extension.py", line 1555, in _run_ninja_build
raise RuntimeError(message) from e
RuntimeError: Error compiling objects for extension

Wrong results on Classification work ? Does PointWeb really work on classfication job?

I tried to build a classification network based on the code in pointweb segment, use the same setting provided by the paper, while what I got on the test case is 8% acc on Training part, even less on testing part

The final results

Thanks for your sharing! I wonder hwo you determine the final results.
Were you using the model trained in the last epoch to test
or test in each epoch and report the best one?

undefined symbol: _ZN6caffe26detail37_typeMetaDataInstance_preallocated_32E

When I try to run sh tool/train.sh s3dis pointweb, I get the following error:

[2020-02-06 23:16:49,255 INFO train.py line 65 10012] arch: pointweb_seg
base_lr: 0.05
block_size: 1.0
classes: 13
data_name: s3dis
data_root: dataset/s3dis
epochs: 100
evaluate: True
fea_dim: 6
ignore_label: 255
manual_seed: None
model_path: exp/s3dis/pointweb/model/train_epoch_100.pth
momentum: 0.9
multiplier: 0.1
names_path: data/s3dis/s3dis_names.txt
num_point: 4096
print_freq: 10
resume: None
sample_rate: 1.0
save_folder: exp/s3dis/pointweb/result/epoch_100/val5_0.5
save_freq: 1
save_path: exp/s3dis/pointweb/model
split: val
start_epoch: 0
step_epoch: 30
stride_rate: 0.5
sync_bn: False
test_area: 5
test_batch_size: 8
test_gpu: [0]
test_list: dataset/s3dis/list/val5.txt
test_list_full: dataset/s3dis/list/val5_full.txt
test_workers: 4
train_batch_size: 16
train_batch_size_val: 8
train_full_folder: dataset/s3dis/trainval_fullarea
train_gpu: [0]
train_list: dataset/s3dis/list/train12346.txt
train_workers: 8
use_xyz: True
val_list: dataset/s3dis/list/val5.txt
weight: None
weight_decay: 0.0001
Traceback (most recent call last):
  File "tool/train.py", line 274, in <module>
    main()
  File "tool/train.py", line 75, in main
    from model.pointweb.pointweb_seg import PointWebSeg as Model
  File "/home/niko/workspace/PointWeb/model/pointweb/pointweb_seg.py", line 6, in <module>
    from model.pointweb.pointweb_module import PointWebSAModule
  File "/home/niko/workspace/PointWeb/model/pointweb/pointweb_module.py", line 7, in <module>
    from lib.pointops.functions import pointops
  File "/home/niko/workspace/PointWeb/lib/pointops/functions/pointops.py", line 7, in <module>
    import pointops_cuda
ImportError: /home/niko/workspace/PointWeb/venv/lib/python3.6/site-packages/pointops-0.0.0-py3.6-linux-x86_64.egg/pointops_cuda.cpython-36m-x86_64-linux-gnu.so: undefined symbol: _ZN6caffe26detail37_typeMetaDataInstance_preallocated_32E
[2020-02-06 23:16:49,622 INFO test_s3dis.py line 49 10025] arch: pointweb_seg
base_lr: 0.05
block_size: 1.0
classes: 13
data_name: s3dis
data_root: dataset/s3dis
epochs: 100
evaluate: True
fea_dim: 6
ignore_label: 255
manual_seed: None
model_path: exp/s3dis/pointweb/model/train_epoch_100.pth
momentum: 0.9
multiplier: 0.1
names_path: data/s3dis/s3dis_names.txt
num_point: 4096
print_freq: 10
resume: None
sample_rate: 1.0
save_folder: exp/s3dis/pointweb/result/epoch_100/val5_0.5
save_freq: 1
save_path: exp/s3dis/pointweb/model
split: val
start_epoch: 0
step_epoch: 30
stride_rate: 0.5
sync_bn: True
test_area: 5
test_batch_size: 8
test_gpu: [0]
test_list: dataset/s3dis/list/val5.txt
test_list_full: dataset/s3dis/list/val5_full.txt
test_workers: 4
train_batch_size: 16
train_batch_size_val: 8
train_full_folder: dataset/s3dis/trainval_fullarea
train_gpu: [0]
train_list: dataset/s3dis/list/train12346.txt
train_workers: 8
use_xyz: True
val_list: dataset/s3dis/list/val5.txt
weight: None
weight_decay: 0.0001
[2020-02-06 23:16:49,622 INFO test_s3dis.py line 51 10025] => creating model ...
[2020-02-06 23:16:49,622 INFO test_s3dis.py line 52 10025] Classes: 13
Traceback (most recent call last):
  File "tool/test_s3dis.py", line 216, in <module>
    main()
  File "tool/test_s3dis.py", line 59, in main
    from model.pointweb.pointweb_seg import PointWebSeg as Model
  File "/home/niko/workspace/PointWeb/model/pointweb/pointweb_seg.py", line 6, in <module>
    from model.pointweb.pointweb_module import PointWebSAModule
  File "/home/niko/workspace/PointWeb/model/pointweb/pointweb_module.py", line 7, in <module>
    from lib.pointops.functions import pointops
  File "/home/niko/workspace/PointWeb/lib/pointops/functions/pointops.py", line 7, in <module>
    import pointops_cuda
ImportError: /home/niko/workspace/PointWeb/venv/lib/python3.6/site-packages/pointops-0.0.0-py3.6-linux-x86_64.egg/pointops_cuda.cpython-36m-x86_64-linux-gnu.so: undefined symbol: _ZN6caffe26detail37_typeMetaDataInstance_preallocated_32E

just remind: the link of your paper in readme.md is invalid

Training with 2 GPUs

Hi, I wish to know is it possible to train PointWeb on Scannet with only 2 GPUs (RTX 2080, each with ~8000MB memory)?

Should people make changes somewhere in the code? If so, could you give some hints about how?

Many thanks!

Can you provide code for the Modelnet40 classification model?

I modified it according to the method mentioned in the paper，but get a bad result.
could you provide the code of classification model which accuracy reach 92%

x86_64-linux-gnu-gcc'failed with exit status 1

Hi,sorry to bother you.When i used the command“ python setup.py install”,I encontered warnings like this:Tensor.data() is deprecated,and lastly met the error:command'x86_64-linux-gnu-gcc'failed with exit status 1".Hoping your answer,thanks very much.

Recommend Projects

React

A declarative, efficient, and flexible JavaScript library for building user interfaces.
Vue.js

🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
Typescript

TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
TensorFlow

An Open Source Machine Learning Framework for Everyone
Django

The Web framework for perfectionists with deadlines.
Laravel

A PHP framework for web artisans
D3

Bring data to life with SVG, Canvas and HTML. 📊📈🎉

Recommend Topics

javascript

JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
web

Some thing interesting about web. New door for the world.
server

A server is a program made to process requests and deliver data to clients.
Machine learning

Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Visualization

Some thing interesting about visualization, use data art
Game

Some thing interesting about game, make everyone happy.

Recommend Org

Facebook

We are working to build community through open source technology. NB: members must have two-factor auth.
Microsoft

Open source projects and samples from Microsoft.
Google

Google ❤️ Open Source for everyone.
Alibaba

Alibaba Open Source for everyone
D3

Data-Driven Documents codes.
Tencent

China tencent open source team.