hiroharu-kato / mesh_reconstruction Goto Github PK

Hi,

Thanks for open-sourcing this awesome project. I'm trying to implement the mesh reconstruction in Pytorch. So far I can get some decent results on the car category except the obvious artifacts shown below:

After some exploration, I found this seems to be relevant to the viewing_angle, which is set to 15 in the experiment. I am a newbie in 3D graphics, I would appreciate if you could shed some light on this parameter and possible reasons for the artifacts shown above.

Hi,

Mesh_construction_model.zip file is not downloading. The link has been freezed. Kindly provide new link for the file.
this is the link which I am unable to download
https://drive.google.com/open?id=1tRHQoc0VWpj61PM1tVozIFwrFsDpKbTQ

thanks

Hi, When I run the reconstruct.py,it show me this,I changed the GPU=0 to try some other gpus,it still doesn't work and show me that.

Traceback (most recent call last): File "mesh_reconstruction/reconstruct.py", line 81, in <module> run() File "mesh_reconstruction/reconstruct.py", line 49, in run cp.random.seed(args.seed) File "/usr/local/lib/python2.7/dist-packages/cupy/random/generator.py", line 1113, in seed get_random_state().seed(seed) File "/usr/local/lib/python2.7/dist-packages/cupy/random/generator.py", line 1147, in get_random_state rs = RandomState(seed) File "/usr/local/lib/python2.7/dist-packages/cupy/random/generator.py", line 51, in __init__ self._generator = curand.createGenerator(method) File "cupy/cuda/curand.pyx", line 90, in cupy.cuda.curand.createGenerator File "cupy/cuda/curand.pyx", line 94, in cupy.cuda.curand.createGenerator File "cupy/cuda/curand.pyx", line 83, in cupy.cuda.curand.check_status cupy.cuda.curand.CURANDError: CURAND_STATUS_INITIALIZATION_FAILED

I have a try to comment cp.random.seed(args.seed)，and it show me another infomation:
(I changed the code : GPU=2 in line 14 and #cp.random.seed(args.seed))

Traceback (most recent call last): File "mesh_reconstruction/reconstruct.py", line 81, in <module> run() File "mesh_reconstruction/reconstruct.py", line 61, in run model.to_gpu() File "/mnt/mesh_reconstruction-master/mesh_reconstruction/models.py", line 112, in to_gpu super(Model, self).to_gpu() File "/usr/local/lib/python2.7/dist-packages/chainer/device_resident.py", line 104, in to_gpu self.__to_device(visitor) File "/usr/local/lib/python2.7/dist-packages/chainer/device_resident.py", line 149, in __to_device self.device_resident_accept(to_device_visitor) File "/usr/local/lib/python2.7/dist-packages/chainer/link.py", line 961, in device_resident_accept d[name].device_resident_accept(visitor) File "/usr/local/lib/python2.7/dist-packages/chainer/link.py", line 958, in device_resident_accept super(Chain, self).device_resident_accept(visitor) File "/usr/local/lib/python2.7/dist-packages/chainer/link.py", line 447, in device_resident_accept super(Link, self).device_resident_accept(visitor) File "/usr/local/lib/python2.7/dist-packages/chainer/device_resident.py", line 38, in device_resident_accept visitor.visit_device_resident(self) File "/usr/local/lib/python2.7/dist-packages/chainer/device_resident.py", line 269, in visit_device_resident to_method(**kwargs) TypeError: to_gpu() got an unexpected keyword argument 'device'

Hi @hiroharu-kato
Thanks for the code. I am testing it for my further research. Can you please provide the requirements.txt

Hi @hiroharu-kato ,Thanks for your excellent work. I am wondering that how does Decoder work. I didn't get it's details in your paper and it's hard to understand the code without enough comments.

Hello!

I'm learning your repo but I found no code for generating the view images from the ShapeNetCore dataset. You mentioned in your paper that you used Blender and rendered from 24 azimuth angles with a fixed elevation angle but I still wonder if you can release the code of this process.
I want to produce the input images under the same settings for lighting and camera etc., on a different train/val/test division.

Thanks.

Hi, when I run python setup.py install --user
it shows :
Traceback (most recent call last):
File "setup.py", line 3, in
import neural_renderer
File "D:\Single-Image-3D-Reconstruction-Neural-Renderer\mesh_reconstruction\neural_renderer\neural_renderer_init_.py", line 1, in
from cross import cross
ModuleNotFoundError: No module named 'cross'
How can I do?

HI @hiroharu-kato

I am following your model code however i have tried to run the command
python mesh_reconstruction/reconstruct.py -d ./data/models -eid singleclass_02691156 -i ./data/examples/airplane_in.png -oi ./data/examples/airplane_out.png -oo ./data/examples/airplane_out.obj but after running this code returns an error called
TypeError: to_gpu() got an unexpected keyword argument 'device'
kindly suggest what to do to make it run

thanks

Hi!!

This is such an amazing repo! And I have two questions hope you could help me with :)

1. the current network supports higher quality rendering? i.e. In the Texture Optimizing the texture is the size of 256 x 256 picture which doesn't hold much detail.
   If I want to render with a higher quality texture, for example a 1080 x 1080 texture, I will have to down resolution to 256 x 256 to fit the network, but the resolution quality of the final output is really low, do you minding letting how or if I could modify the network to work with higher resolution texture.

2. In your examples, all the output were exported as gif, I'm wondering if I can export it as obj files?

Thank you so much, and have a nice day 😃

~/mesh_reconstruction-master$ python mesh_reconstruction/train.py -eid singleclass_02691156 -cls 02691156 -ls 0.001 -li 1000 -ni 200000
Loading dataset: 100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 1/1 [00:03<00:00, 3.54s/it]
Loading dataset: 100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 1/1 [00:00<00:00, 1.88it/s]
iteration main/loss_silhouettes main/loss_smoothness val/iou elapsed_time
Exception in main training loop: /tmp/tmpbY1edp/bf28e852c345f19f0f38495ee3c20355_2.cubin.cu(19): error: a value of type "const float *" cannot be used to initialize an entity of type "float *"
this epoch [..................................................] 0.00%
1 error detected in the compilation of "/tmp/tmpbY1edp/bf28e852c345f19f0f38495ee3c20355_2.cubin.cu".
7.3165 iters/sec. Estimated time to finish: 7:33:20.194782.
Traceback (most recent call last):
File "/home/sm/.local/lib/python2.7/site-packages/chainer/training/trainer.py", line 319, in run
entry.extension(self)
File "/home/sm/mesh_reconstruction-master/mesh_reconstruction/training.py", line 37, in validation
iou += model.evaluate_iou(*batch).sum()
File "/home/sm/mesh_reconstruction-master/mesh_reconstruction/models.py", line 149, in evaluate_iou
voxels_predicted = voxelization.voxelize(faces, 32, False)
File "/home/sm/mesh_reconstruction-master/mesh_reconstruction/voxelization.py", line 175, in voxelize
voxels0 = _voxelize_sub1(faces, size, 0)
File "/home/sm/mesh_reconstruction-master/mesh_reconstruction/voxelization.py", line 64, in _voxelize_sub1
)(cp.arange(bs * size * size).astype('int32'), faces, bs, nf, size, voxels)
File "cupy/core/_kernel.pyx", line 572, in cupy.core._kernel.ElementwiseKernel.call
File "cupy/core/_kernel.pyx", line 602, in cupy.core._kernel.ElementwiseKernel._get_elementwise_kernel
File "cupy/core/_kernel.pyx", line 402, in cupy.core._kernel._get_elementwise_kernel
File "cupy/core/_kernel.pyx", line 40, in cupy.core._kernel._get_simple_elementwise_kernel
File "cupy/core/_kernel.pyx", line 60, in cupy.core._kernel._get_simple_elementwise_kernel
File "cupy/core/carray.pxi", line 166, in cupy.core.core.compile_with_cache
File "/home/sm/.conda/envs/NR1/lib/python2.7/site-packages/cupy/cuda/compiler.py", line 166, in compile_with_cache
ptx = compile_using_nvrtc(source, options, arch, name + '.cu')
File "/home/sm/.conda/envs/NR1/lib/python2.7/site-packages/cupy/cuda/compiler.py", line 82, in compile_using_nvrtc
ptx = prog.compile(options)
File "/home/sm/.conda/envs/NR1/lib/python2.7/site-packages/cupy/cuda/compiler.py", line 253, in compile
raise CompileException(log, self.src, self.name, options)
Will finalize trainer extensions and updater before reraising the exception.
Traceback (most recent call last):
File "mesh_reconstruction/train.py", line 94, in
run()
File "mesh_reconstruction/train.py", line 85, in run
trainer.run()
File "/home/sm/.local/lib/python2.7/site-packages/chainer/training/trainer.py", line 330, in run
six.reraise(*sys.exc_info())
File "/home/sm/.local/lib/python2.7/site-packages/chainer/training/trainer.py", line 319, in run
entry.extension(self)
File "/home/sm/mesh_reconstruction-master/mesh_reconstruction/training.py", line 37, in validation
iou += model.evaluate_iou(*batch).sum()
File "/home/sm/mesh_reconstruction-master/mesh_reconstruction/models.py", line 149, in evaluate_iou
voxels_predicted = voxelization.voxelize(faces, 32, False)
File "/home/sm/mesh_reconstruction-master/mesh_reconstruction/voxelization.py", line 175, in voxelize
voxels0 = _voxelize_sub1(faces, size, 0)
File "/home/sm/mesh_reconstruction-master/mesh_reconstruction/voxelization.py", line 64, in _voxelize_sub1
)(cp.arange(bs * size * size).astype('int32'), faces, bs, nf, size, voxels)
File "cupy/core/_kernel.pyx", line 572, in cupy.core._kernel.ElementwiseKernel.call
File "cupy/core/_kernel.pyx", line 602, in cupy.core._kernel.ElementwiseKernel._get_elementwise_kernel
File "cupy/core/_kernel.pyx", line 402, in cupy.core._kernel._get_elementwise_kernel
File "cupy/core/_kernel.pyx", line 40, in cupy.core._kernel._get_simple_elementwise_kernel
File "cupy/core/_kernel.pyx", line 60, in cupy.core._kernel._get_simple_elementwise_kernel
File "cupy/core/carray.pxi", line 166, in cupy.core.core.compile_with_cache
File "/home/sm/.conda/envs/NR1/lib/python2.7/site-packages/cupy/cuda/compiler.py", line 166, in compile_with_cache
ptx = compile_using_nvrtc(source, options, arch, name + '.cu')
File "/home/sm/.conda/envs/NR1/lib/python2.7/site-packages/cupy/cuda/compiler.py", line 82, in compile_using_nvrtc
ptx = prog.compile(options)
File "/home/sm/.conda/envs/NR1/lib/python2.7/site-packages/cupy/cuda/compiler.py", line 253, in compile
raise CompileException(log, self.src, self.name, options)
cupy.cuda.compiler.CompileException: /tmp/tmpbY1edp/bf28e852c345f19f0f38495ee3c20355_2.cubin.cu(19): error: a value of type "const float *" cannot be used to initialize an entity of type "float *"

1 error detected in the compilation of "/tmp/tmpbY1edp/bf28e852c345f19f0f38495ee3c20355_2.cubin.cu".