zhengzerong / pamir Goto Github PK

Hi,
when I run “python main_calc_prt.py”, code in mesh.ray.intersects_any() doesn't respond. Every one have any suggestions on that?

Hi Zerong

Could you provide some references to show how to generate these two files from SMPL model?

tetrahedron.txt

tetra_smpl.npy

Thanks a lot

when you train models, how many views you use for one model(each 1 degree, or each 10 degree or others?)

I am keep getting no module named 'voxelize_cuda' error when run python main_test.py.
I am using pytorch 1.7.0
Cuda 10.1
Could anyone tell me where could possible be wrong? Thx

Using /home/sijia/.cache/torch_extensions as PyTorch extensions root...
Detected CUDA files, patching ldflags
Emitting ninja build file /home/sijia/.cache/torch_extensions/voxelize_cuda/build.ninja...
Building extension module voxelize_cuda...
Allowing ninja to set a default number of workers... (overridable by setting the environment variable MAX_JOBS=N)
1.10.2
Loading extension module voxelize_cuda...
Traceback (most recent call last):
File "main_test.py", line 133, in
main_test_wo_gt_smpl_with_optm(input_image_dir,
File "main_test.py", line 36, in main_test_wo_gt_smpl_with_optm
from evaluator import Evaluator
File "/home/sijia/Desktop/PaMIR/networks/evaluator.py", line 23, in
from neural_voxelization_layer.voxelize import Voxelization
File "/home/sijia/Desktop/PaMIR/networks/neural_voxelization_layer/voxelize.py", line 13, in
voxelize_cuda = load(
File "/home/sijia/anaconda3/envs/pamir/lib/python3.8/site-packages/torch/utils/cpp_extension.py", line 969, in load
return _jit_compile(
File "/home/sijia/anaconda3/envs/pamir/lib/python3.8/site-packages/torch/utils/cpp_extension.py", line 1196, in _jit_compile
return _import_module_from_library(name, build_directory, is_python_module)
File "/home/sijia/anaconda3/envs/pamir/lib/python3.8/site-packages/torch/utils/cpp_extension.py", line 1543, in _import_module_from_library
file, path, description = imp.find_module(module_name, [path])
File "/home/sijia/anaconda3/envs/pamir/lib/python3.8/imp.py", line 296, in find_module
raise ImportError(_ERR_MSG.format(name), name=name)
ImportError: No module named 'voxelize_cuda'
(pamir) sijia@sijia-750-170se:~/Desktop/PaMIR/networks$ python3 main_test.py
True
Using /home/sijia/.cache/torch_extensions as PyTorch extensions root...
Detected CUDA files, patching ldflags
Emitting ninja build file /home/sijia/.cache/torch_extensions/voxelize_cuda/build.ninja...
Building extension module voxelize_cuda...
Allowing ninja to set a default number of workers... (overridable by setting the environment variable MAX_JOBS=N)
1.10.2
Loading extension module voxelize_cuda...
Traceback (most recent call last):
File "main_test.py", line 133, in
main_test_wo_gt_smpl_with_optm(input_image_dir,
File "main_test.py", line 36, in main_test_wo_gt_smpl_with_optm
from evaluator import Evaluator
File "/home/sijia/Desktop/PaMIR/networks/evaluator.py", line 23, in
from neural_voxelization_layer.voxelize import Voxelization
File "/home/sijia/Desktop/PaMIR/networks/neural_voxelization_layer/voxelize.py", line 13, in
voxelize_cuda = load(
File "/home/sijia/anaconda3/envs/pamir/lib/python3.8/site-packages/torch/utils/cpp_extension.py", line 969, in load
return _jit_compile(
File "/home/sijia/anaconda3/envs/pamir/lib/python3.8/site-packages/torch/utils/cpp_extension.py", line 1196, in _jit_compile
return _import_module_from_library(name, build_directory, is_python_module)
File "/home/sijia/anaconda3/envs/pamir/lib/python3.8/site-packages/torch/utils/cpp_extension.py", line 1543, in _import_module_from_library
file, path, description = imp.find_module(module_name, [path])
File "/home/sijia/anaconda3/envs/pamir/lib/python3.8/imp.py", line 296, in find_module
raise ImportError(_ERR_MSG.format(name), name=name)
ImportError: No module named 'voxelize_cuda'

Dear authors,
Thanks for your great work~
I want to know how to convert the RenderPeople mesh to the SMPL model, and whether it is handled by optimizing betas and thetas parameters? Do you have any codes for it ?

When I run the main_sample_occ.py for THuman2.0 dataset, it displays "Mesh is non-watertight for contained point query". Is that normal? How can I fix this problem

My GPU(6GiB) was out of memory, when it doing SDF query on the demo. Is there anything I can do to reduce the memory occupation to run the demo?

FileNotFoundError: [Errno 2] No such file or directory: '../dataset_example/mesh_data/./0000/bounce/prt_data.mat'

Thank you for your work. This is an interesting project.
But when I ran your code, I encountered this error:
RuntimeError: CUDA error: all CUDA-capable devices are busy or unavailable.
But running other codes in the same environment will not cause this error.
Have you ever encountered similar situations, and can you tell me how to solve them

Run python main_render_images.py gave this error.

(PaMIR) yyyfan@yyyfan-MS-7A63:~/PaMIR/data$ python main_render_images.py
0%| | 0/1 [00:00<?, ?it/sSegmentation fault (core dumped) | 0/360 [00:00<?, ?it/s]

I found it was caused by line 332 in lib/render/gl/prt_render.py
glDrawArrays(GL_TRIANGLES, 0, int(self.n_vertices[mat]))
I test on two machines and it gives the same error. Does anyone met the same error before?

Thank you so much for this awesome work! I am running the demo with your code. I noticed during the test, I need to make an input mask image, but I can't find the code to generate it.

Could you share the code or any reference so I can work same with you?

I'm sorry to bother you, I ran the demo code with the data provided, and got very strange results like this, but I don't know the reason. Could you please tell me the reason or help me analyze it ?Really thanks!

File "main_test.py", line 138, in
main_test_texture(output_dir,
File "main_test.py", line 100, in main_test_texture
raise FileNotFoundError('Cannot found SMPL parameters! You need to run PaMIR-geometry first!')
FileNotFoundError: Cannot found SMPL parameters! You need to run PaMIR-geometry first!

Sorry to bother you, after training the model, several test images in the original test data file can be predicted, but when I add new pictures, such a problem occurs, but the remaining pictures can be tested correctly. When I take this trained model to another computer for testing, the new images added to the test data folder can be pre-tested. After many comparisons, I found that such a problem does exist, but I don't know the reason, so I came here to ask for advice. What is the possible reason?

For your paper under implementation details it says you use 900 scans from twindom and 600 scans from deephuman. Then it says you train for 9 epochs with a batch size of 3. If I'm not mistaken that would be 500 steps per batch right? You say that every 10,000 iterations you reduce the learning rate but Im not sure how you get above 4,500 steps?

Also from your code it seems you don't normalize the input image for the hourglass network. Is there any particular reason for that?

Hi! Thanks for contributing such a great work and dataset. I just wonder is the THuman2.0 dataset large enough to train a model like PaMIR for the single view human reconstruction task. Can you share some results?

Thanks for your great work~
In the paper, it is mentioned that GCMR has been re-trained by your dataset, can you provide any specific details about it?

anyone know what the python version is? thank you very much.

Can you please advice is pre-trained checkpoint compatible with smpl_men_model? or it can be used only for neutral model?

Thanks for your releasing codes! I have some confusion about your article. There are two innovative points in proposed method, Reconstruction Loss and Body Reference Optimization. However, I do not understand what the phrase"image observation" means in part Body Reference Optimization. Could you explain it a little clearly? Thanks a lot!

用你们的采样点，训练的PIFu loss 最终收敛在0.3， IOU 大约在0.68，但是用原始的PIFu的采样方式切可以达到0.83.我不知道在什么环节上除了错误，需要你的帮助- -