resscnn's People
Contributors
Stargazers
Watchers
Forkers
liuhahahaharesscnn's Issues
Dataset Issue
Hi
The following dataset after downloading returns an empty zip;
Link for whole dataset with generated pseudo MOS
Also, in the following table from the paper, you mentioned these three datasets,
Would you please indicate their related links?
I appreciate it
Train Loss: nan
All the epochs of training return Nan as the training loss.
I checked it and it is not coming from the nan gradient.
Still not sure what causes this issue!
2022-12-06 12:43:12,673 - /home/ResSCNN/lib/Logging.py[line:25] - INFO: Final Loss: 3.499,
2022-12-06 12:43:12,674 - /home/ResSCNN/lib/Logging.py[line:25] - INFO: Test Loss: 3.499211
2022-12-06 12:43:12,701 - /home/ResSCNN/lib/Logging.py[line:25] - INFO: PLCC: 0.997658, SROCC: 0.900000, KROCC: 0.800000, RMSE: 0.000064,
2022-12-06 12:43:12,716 - /home/ResSCNN/lib/Logging.py[line:25] - INFO: Epoch: 1, LR: [0.001]
2022-12-06 12:43:12,951 - /home/ResSCNN/lib/Logging.py[line:25] - INFO: Train Epoch: 1 , Train Loss: nan
2022-12-06 12:43:32,335 - /home/ResSCNN/lib/Logging.py[line:25] - INFO: Final Loss: 3.499,
2022-12-06 12:43:32,336 - /home/ResSCNN/lib/Logging.py[line:25] - INFO: Test Loss: 3.499211
2022-12-06 12:43:32,358 - /home/ResSCNN/lib/Logging.py[line:25] - INFO: PLCC: 0.997658, SROCC: 0.900000, KROCC: 0.800000, RMSE: 0.000064,
2022-12-06 12:43:32,383 - /home/ResSCNN/lib/Logging.py[line:25] - INFO: Epoch: 2, LR: [0.0009801]
2022-12-06 12:43:32,648 - /home/ResSCNN/lib/Logging.py[line:25] - INFO: Train Epoch: 2 , Train Loss: nan
2022-12-06 12:43:52,009 - /home/ResSCNN/lib/Logging.py[line:25] - INFO: Final Loss: 3.499,
2022-12-06 12:43:52,012 - /home/ResSCNN/lib/Logging.py[line:25] - INFO: Test Loss: 3.499211
2022-12-06 12:43:52,035 - /home/ResSCNN/lib/Logging.py[line:25] - INFO: PLCC: 0.997658, SROCC: 0.900000, KROCC: 0.800000, RMSE: 0.000064,
Can we train this on our own data
Would we need to generate a MOS score for our own data or can we substitute in our own .ply files? Thanks!
LS-PCQA数据使用问题
您好，我在阅读文章时，对于LS-PCQA数据集的使用有一些疑惑。
在5.1.1节中，您提到 “To evaluate the performance of the proposed ResSCNN, we conduct evaluation experiments using Part III of the proposed LS-PCQA dataset, where the distorted samples are all labeled by pseudo MOS” ，即仅使用数据集的第三部分进行训练和测试。
而在5.1.2节中，您写到 “The training set and testing set from LS-PCQA contain the distorted samples generated from 100 reference point clouds and 4 reference point clouds respectively to avoid overlapping”，这里说的是100个用于训练，4个用于测试，一共使用了104个原始点云的失真版本，即使用整个LS=PCQA数据集来训练和测试。
我想向您确认，您是如何使用LS-PCQA数据集的。是用数据集的part III还是整个数据集？以及4个用于测试的点云具体是哪四个？
期待您的回复！
AssertionError: The MinkowskiEngine was compiled with CPU_ONLY flag
When I am compiling the train.py, it returns the following error.
When I was installing MinkowskiEngine I used the pip that in the setup.py there is torch.cuda.is_available() by default!
I tried to move the method to cuda but did not work
/home/saman/anaconda3/envs/py37/lib/python3.7/site-packages/MinkowskiEngine/__init__.py:225: UserWarning: The MinkowskiEngine was compiled with CPU_ONLY flag. If you want to compile with CUDA support, make sure `torch.cuda.is_available()` is True when you install MinkowskiEngine.
"If you want to compile with CUDA support, make sure `torch.cuda.is_available()` is True when you install MinkowskiEngine.",
2022-11-22 11:03:47,975 - /home/saman/ResSCNN/lib/Logging.py[line:25] - INFO: ResSCNN(
(conv1): MinkowskiConvolution(in=3, out=64, kernel_size=[3, 3, 3], stride=[2, 2, 2], dilation=[1, 1, 1])
(norm1): MinkowskiBatchNorm(64, eps=1e-05, momentum=0.05, affine=True, track_running_stats=True)
(block1): BasicBlock(
(conv1): MinkowskiConvolution(in=64, out=64, kernel_size=[3, 3, 3], stride=[1, 1, 1], dilation=[1, 1, 1])
(norm1): MinkowskiBatchNorm(64, eps=1e-05, momentum=0.05, affine=True, track_running_stats=True)
(conv2): MinkowskiConvolution(in=64, out=64, kernel_size=[3, 3, 3], stride=[1, 1, 1], dilation=[1, 1, 1])
(norm2): MinkowskiBatchNorm(64, eps=1e-05, momentum=0.05, affine=True, track_running_stats=True)
)
(conv2): MinkowskiConvolution(in=64, out=64, kernel_size=[3, 3, 3], stride=[2, 2, 2], dilation=[1, 1, 1])
(norm2): MinkowskiBatchNorm(64, eps=1e-05, momentum=0.05, affine=True, track_running_stats=True)
(block2): BasicBlock(
(conv1): MinkowskiConvolution(in=64, out=64, kernel_size=[3, 3, 3], stride=[1, 1, 1], dilation=[1, 1, 1])
(norm1): MinkowskiBatchNorm(64, eps=1e-05, momentum=0.05, affine=True, track_running_stats=True)
(conv2): MinkowskiConvolution(in=64, out=64, kernel_size=[3, 3, 3], stride=[1, 1, 1], dilation=[1, 1, 1])
(norm2): MinkowskiBatchNorm(64, eps=1e-05, momentum=0.05, affine=True, track_running_stats=True)
)
(conv3): MinkowskiConvolution(in=64, out=64, kernel_size=[3, 3, 3], stride=[2, 2, 2], dilation=[1, 1, 1])
(norm3): MinkowskiBatchNorm(64, eps=1e-05, momentum=0.05, affine=True, track_running_stats=True)
(block3): BasicBlock(
(conv1): MinkowskiConvolution(in=64, out=64, kernel_size=[3, 3, 3], stride=[1, 1, 1], dilation=[1, 1, 1])
(norm1): MinkowskiBatchNorm(64, eps=1e-05, momentum=0.05, affine=True, track_running_stats=True)
(conv2): MinkowskiConvolution(in=64, out=64, kernel_size=[3, 3, 3], stride=[1, 1, 1], dilation=[1, 1, 1])
(norm2): MinkowskiBatchNorm(64, eps=1e-05, momentum=0.05, affine=True, track_running_stats=True)
)
(conv4): MinkowskiConvolution(in=64, out=64, kernel_size=[3, 3, 3], stride=[2, 2, 2], dilation=[1, 1, 1])
(norm4): MinkowskiBatchNorm(64, eps=1e-05, momentum=0.05, affine=True, track_running_stats=True)
(block4): BasicBlock(
(conv1): MinkowskiConvolution(in=64, out=64, kernel_size=[3, 3, 3], stride=[1, 1, 1], dilation=[1, 1, 1])
(norm1): MinkowskiBatchNorm(64, eps=1e-05, momentum=0.05, affine=True, track_running_stats=True)
(conv2): MinkowskiConvolution(in=64, out=64, kernel_size=[3, 3, 3], stride=[1, 1, 1], dilation=[1, 1, 1])
(norm2): MinkowskiBatchNorm(64, eps=1e-05, momentum=0.05, affine=True, track_running_stats=True)
)
(glob_avg): AdaptiveMaxPool2d(output_size=(1, None))
(fc1): Linear(in_features=256, out_features=32, bias=True)
(fc2): Linear(in_features=32, out_features=1, bias=True)
)
0%| | 0/5 [00:00<?, ?it/s]
Traceback (most recent call last):
File "/home/saman/anaconda3/envs/py37/lib/python3.7/runpy.py", line 193, in _run_module_as_main
"__main__", mod_spec)
File "/home/saman/anaconda3/envs/py37/lib/python3.7/runpy.py", line 85, in _run_code
exec(code, run_globals)
File "/home/saman/ResSCNN/train.py", line 55, in <module>
main(config)
File "/home/saman/ResSCNN/train.py", line 47, in main
trainer.train()
File "/home/saman/ResSCNN/lib/trainer.py", line 111, in train
val_dict,plcc,srocc = self._test_epoch(0)
File "/home/saman/ResSCNN/lib/trainer.py", line 247, in _test_epoch
stensor = ME.SparseTensor(feats[0].to(self.device), coordinates=coords[0].to(self.device))
File "/home/saman/anaconda3/envs/py37/lib/python3.7/site-packages/MinkowskiEngine/MinkowskiSparseTensor.py", line 252, in __init__
minkowski_algorithm=minkowski_algorithm,
File "/home/saman/anaconda3/envs/py37/lib/python3.7/site-packages/MinkowskiEngine/MinkowskiCoordinateManager.py", line 138, in __init__
), "The MinkowskiEngine was compiled with CPU_ONLY flag. If you want to compile with CUDA support, make sure `torch.cuda.is_available()` is True when you install MinkowskiEngine."
AssertionError: The MinkowskiEngine was compiled with CPU_ONLY flag. If you want to compile with CUDA support, make sure `torch.cuda.is_available()` is True when you install MinkowskiEngine.
Minkowski
May I kindly ask how did you install the Minkowski?
Because there are lots of issues regarding the Minkowski installation on its repository, and its installation problem is a common thing among most of the developers!
Thank you in advance
ERROR! Session/line number was not unique in database. History logging moved to new session 881
I have an issue with train.py, when I am running the code it returns the following error, and it crashes;
Do you have any idea how I can solve this issue, please?
Thank you
ERROR! Session/line number was not unique in database. History logging moved to new session 881
_save_checkpoint
I am trying the save the trained model with the _save_checkpoint method. But it needs the path to dump the model, so I defined the path in config.resume, but it tells me that there is no checkpoint found. Meaning that it needs some dumped model already but it does not start to train to dump the model!
Data shape and size
May I kindly ask about the input size of the model?
I assume you are using a 3-Dimensional tensor.
I am unsure if the dimension of the MinkowskiConvolution layer is 2D or 1D.
Would you please help me with this?
Thank you
关于模型的参数量和flop的问题
您好，我在阅读代码和文章的时候有一些理解，不知道正否正确：
- 不同点数量的点云都要先voxel化，处理成[N,6]的维度来进行处理？
- 不同点数量的点云 处理成input之后维度应该也是不一样的，也就是N是不同的？所以这是batchsize设为1的原因吗？
- 我理解下来模型的参数量应该是固定的？我算下来是1.6M，不知道是否正确？
- flops应该是和处理之后的张量维度有关，也就是对于不同的点云，计算量应该是不一样的？
期待您的回复！
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.