hongfz16 / eva3d Goto Github PK
View Code? Open in Web Editor NEW[ICLR 2023 Spotlight] EVA3D: Compositional 3D Human Generation from 2D Image Collections
License: Other
[ICLR 2023 Spotlight] EVA3D: Compositional 3D Human Generation from 2D Image Collections
License: Other
is there any estimated time of arrival for the code release?
What is the smpl_link parameter for duplicating the HF space?
also your huggingface space is down
https://huggingface.co/spaces/hongfz16/EVA3D
Runtime error
00/160M [00:00<?, ?B/s]�[A
0%| | 0.00/160M [00:00<?, ?B/s]�[A
0%| | 0.00/160M [00:00<?, ?B/s]�[ATraceback (most recent call last):
File "app.py", line 42, in download_pretrained_models
download_file(session, eva3d_deepfashion_model)
File "EVA3D/download_models.py", line 81, in download_file
res.raise_for_status()
File "/home/user/.pyenv/versions/3.8.9/lib/python3.8/site-packages/requests/models.py", line 1021, in raise_for_status
raise HTTPError(http_error_msg, response=self)
requests.exceptions.HTTPError: 500 Server Error: Internal Server Error for url: https://drive.google.com/uc?id=1SYPjxnHz3XPRhTarx_Lw8SG_iz16QUMU&confirm=t&uuid=3a51e74f-d6cd-4692-adcd-867a1dde7a18
During handling of the above exception, another exception occurred:
Traceback (most recent call last):
File "app.py", line 56, in <module>
download_pretrained_models()
File "app.py", line 46, in download_pretrained_models
download_file(session, eva3d_deepfashion_model, use_alt_url=True)
File "EVA3D/download_models.py", line 80, in download_file
with session.get(file_url, stream=True) as res:
File "/home/user/.pyenv/versions/3.8.9/lib/python3.8/site-packages/requests/sessions.py", line 600, in get
return self.request("GET", url, **kwargs)
File "/home/user/.pyenv/versions/3.8.9/lib/python3.8/site-packages/requests/sessions.py", line 573, in request
prep = self.prepare_request(req)
File "/home/user/.pyenv/versions/3.8.9/lib/python3.8/site-packages/requests/sessions.py", line 484, in prepare_request
p.prepare(
File "/home/user/.pyenv/versions/3.8.9/lib/python3.8/site-packages/requests/models.py", line 368, in prepare
self.prepare_url(url, params)
File "/home/user/.pyenv/versions/3.8.9/lib/python3.8/site-packages/requests/models.py", line 439, in prepare_url
raise MissingSchema(
requests.exceptions.MissingSchema: Invalid URL '': No scheme supplied. Perhaps you meant https://?
0%| | 0.00/160M [00:00<?, ?B/s]
0%| | 0.00/160M [00:00<?, ?B/s]
Container logs:
Cloning into 'EVA3D'...
Collecting fvcore
Downloading fvcore-0.1.5.post20221221.tar.gz (50 kB)
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 50.2/50.2 kB 721.9 kB/s eta 0:00:00
Preparing metadata (setup.py): started
Preparing metadata (setup.py): finished with status 'done'
Requirement already satisfied: plotly in /home/user/.pyenv/versions/3.8.9/lib/python3.8/site-packages (5.13.1)
Collecting plotly
Downloading plotly-5.15.0-py2.py3-none-any.whl (15.5 MB)
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 15.5/15.5 MB 32.9 MB/s eta 0:00:00
Requirement already satisfied: numpy in /home/user/.pyenv/versions/3.8.9/lib/python3.8/site-packages (from fvcore) (1.21.2)
Collecting yacs>=0.1.6
Downloading yacs-0.1.8-py3-none-any.whl (14 kB)
Requirement already satisfied: pyyaml>=5.1 in /home/user/.pyenv/versions/3.8.9/lib/python3.8/site-packages (from fvcore) (6.0)
Requirement already satisfied: tqdm in /home/user/.pyenv/versions/3.8.9/lib/python3.8/site-packages (from fvcore) (4.64.1)
Collecting termcolor>=1.1
Downloading termcolor-2.3.0-py3-none-any.whl (6.9 kB)
Requirement already satisfied: Pillow in /home/user/.pyenv/versions/3.8.9/lib/python3.8/site-packages (from fvcore) (9.0.1)
Collecting tabulate
Downloading tabulate-0.9.0-py3-none-any.whl (35 kB)
Collecting iopath>=0.1.7
Downloading iopath-0.1.10.tar.gz (42 kB)
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 42.2/42.2 kB 760.8 kB/s eta 0:00:00
Preparing metadata (setup.py): started
Preparing metadata (setup.py): finished with status 'done'
Requirement already satisfied: tenacity>=6.2.0 in /home/user/.pyenv/versions/3.8.9/lib/python3.8/site-packages (from plotly) (8.2.2)
Requirement already satisfied: packaging in /home/user/.pyenv/versions/3.8.9/lib/python3.8/site-packages (from plotly) (23.0)
Requirement already satisfied: typing_extensions in /home/user/.pyenv/versions/3.8.9/lib/python3.8/site-packages (from iopath>=0.1.7->fvcore) (4.5.0)
Collecting portalocker
Downloading portalocker-2.7.0-py2.py3-none-any.whl (15 kB)
Building wheels for collected packages: fvcore, iopath
Building wheel for fvcore (setup.py): started
Building wheel for fvcore (setup.py): finished with status 'done'
Created wheel for fvcore: filename=fvcore-0.1.5.post20221221-py3-none-any.whl size=61405 sha256=585e93d8432e05cf12db25b5afcdc7f126e03ac93b73784ce8d2ca21064f083a
Stored in directory: /home/user/.cache/pip/wheels/6d/1f/39/577cab48487ad5c31acd046ee33dea7ca6fede7e923b2a2bc1
Building wheel for iopath (setup.py): started
Building wheel for iopath (setup.py): finished with status 'done'
Created wheel for iopath: filename=iopath-0.1.10-py3-none-any.whl size=31532 sha256=0b594ceaf50d8e5be928e2a127ebdfd67283cf96ada1d623d808621c3e253cc9
Stored in directory: /home/user/.cache/pip/wheels/85/e6/c8/4a67ea2cd453fd2e02c616b615fe8655ae67246c8e669bd464
Successfully built fvcore iopath
Installing collected packages: yacs, termcolor, tabulate, portalocker, plotly, iopath, fvcore
Attempting uninstall: plotly
Found existing installation: plotly 5.13.1
Uninstalling plotly-5.13.1:
Successfully uninstalled plotly-5.13.1
Successfully installed fvcore-0.1.5.post20221221 iopath-0.1.10 plotly-5.15.0 portalocker-2.7.0 tabulate-0.9.0 termcolor-2.3.0 yacs-0.1.8
[notice] A new release of pip available: 22.3.1 -> 23.1.2
[notice] To update, run: pip install --upgrade pip
Looking in links: https://dl.fbaipublicfiles.com/pytorch3d/packaging/wheels/py38_cu116_pyt1131/download.html
Collecting pytorch3d
Downloading https://dl.fbaipublicfiles.com/pytorch3d/packaging/wheels/py38_cu116_pyt1131/pytorch3d-0.7.2-cp38-cp38-linux_x86_64.whl (72.8 MB)
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 72.8/72.8 MB 258.8 MB/s eta 0:00:00
Requirement already satisfied: iopath in /home/user/.pyenv/versions/3.8.9/lib/python3.8/site-packages (from pytorch3d) (0.1.10)
Requirement already satisfied: fvcore in /home/user/.pyenv/versions/3.8.9/lib/python3.8/site-packages (from pytorch3d) (0.1.5.post20221221)
Requirement already satisfied: tabulate in /home/user/.pyenv/versions/3.8.9/lib/python3.8/site-packages (from fvcore->pytorch3d) (0.9.0)
Requirement already satisfied: termcolor>=1.1 in /home/user/.pyenv/versions/3.8.9/lib/python3.8/site-packages (from fvcore->pytorch3d) (2.3.0)
Requirement already satisfied: pyyaml>=5.1 in /home/user/.pyenv/versions/3.8.9/lib/python3.8/site-packages (from fvcore->pytorch3d) (6.0)
Requirement already satisfied: yacs>=0.1.6 in /home/user/.pyenv/versions/3.8.9/lib/python3.8/site-packages (from fvcore->pytorch3d) (0.1.8)
Requirement already satisfied: tqdm in /home/user/.pyenv/versions/3.8.9/lib/python3.8/site-packages (from fvcore->pytorch3d) (4.64.1)
Requirement already satisfied: numpy in /home/user/.pyenv/versions/3.8.9/lib/python3.8/site-packages (from fvcore->pytorch3d) (1.21.2)
Requirement already satisfied: Pillow in /home/user/.pyenv/versions/3.8.9/lib/python3.8/site-packages (from fvcore->pytorch3d) (9.0.1)
Requirement already satisfied: portalocker in /home/user/.pyenv/versions/3.8.9/lib/python3.8/site-packages (from iopath->pytorch3d) (2.7.0)
Requirement already satisfied: typing-extensions in /home/user/.pyenv/versions/3.8.9/lib/python3.8/site-packages (from iopath->pytorch3d) (4.5.0)
Installing collected packages: pytorch3d
Successfully installed pytorch3d-0.7.2
Downloading EVA3D model pretrained on DeepFashion.
Google Drive download failed.
Trying do download from alternate server
0%| | 0.00/160M [00:00<?, ?B/s]
0%| | 0.00/160M [00:00<?, ?B/s]
0%| | 2.26k/160M [00:00<3:26:04, 13.0kB/s]
0%| | 0.00/160M [00:00<?, ?B/s]
0%| | 0.00/160M [00:00<?, ?B/s]
0%| | 0.00/160M [00:00<?, ?B/s]
0%| | 0.00/160M [00:00<?, ?B/s]
0%| | 0.00/160M [00:00<?, ?B/s]
0%| | 0.00/160M [00:00<?, ?B/s]
0%| | 0.00/160M [00:00<?, ?B/s]
0%| | 0.00/160M [00:00<?, ?B/s]
0%| | 0.00/160M [00:00<?, ?B/s]
0%| | 0.00/160M [00:00<?, ?B/s]
0%| | 0.00/160M [00:00<?, ?B/s]
0%| | 0.00/160M [00:00<?, ?B/s]
0%| | 0.00/160M [00:00<?, ?B/s]
0%| | 0.00/160M [00:00<?, ?B/s]
0%| | 0.00/160M [00:00<?, ?B/s]
0%| | 0.00/160M [00:00<?, ?B/s]
0%| | 0.00/160M [00:00<?, ?B/s]
0%| | 0.00/160M [00:00<?, ?B/s]
0%| | 0.00/160M [00:00<?, ?B/s]
0%| | 0.00/160M [00:00<?, ?B/s]Traceback (most recent call last):
File "app.py", line 42, in download_pretrained_models
download_file(session, eva3d_deepfashion_model)
File "EVA3D/download_models.py", line 81, in download_file
res.raise_for_status()
File "/home/user/.pyenv/versions/3.8.9/lib/python3.8/site-packages/requests/models.py", line 1021, in raise_for_status
raise HTTPError(http_error_msg, response=self)
requests.exceptions.HTTPError: 500 Server Error: Internal Server Error for url: https://drive.google.com/uc?id=1SYPjxnHz3XPRhTarx_Lw8SG_iz16QUMU&confirm=t&uuid=3a51e74f-d6cd-4692-adcd-867a1dde7a18
During handling of the above exception, another exception occurred:
Traceback (most recent call last):
File "app.py", line 56, in <module>
download_pretrained_models()
File "app.py", line 46, in download_pretrained_models
download_file(session, eva3d_deepfashion_model, use_alt_url=True)
File "EVA3D/download_models.py", line 80, in download_file
with session.get(file_url, stream=True) as res:
File "/home/user/.pyenv/versions/3.8.9/lib/python3.8/site-packages/requests/sessions.py", line 600, in get
return self.request("GET", url, **kwargs)
File "/home/user/.pyenv/versions/3.8.9/lib/python3.8/site-packages/requests/sessions.py", line 573, in request
prep = self.prepare_request(req)
File "/home/user/.pyenv/versions/3.8.9/lib/python3.8/site-packages/requests/sessions.py", line 484, in prepare_request
p.prepare(
File "/home/user/.pyenv/versions/3.8.9/lib/python3.8/site-packages/requests/models.py", line 368, in prepare
self.prepare_url(url, params)
File "/home/user/.pyenv/versions/3.8.9/lib/python3.8/site-packages/requests/models.py", line 439, in prepare_url
raise MissingSchema(
requests.exceptions.MissingSchema: Invalid URL '': No scheme supplied. Perhaps you meant https://?
0%| | 0.00/160M [00:00<?, ?B/s]
0%| | 0.00/160M [00:00<?, ?B/s]
Great project.
I was able to make it work here on my windows and I've got a doubt.
The way it is, is possible to create using custom images?
From what I saw on the code, seems like it can create (at the moment) using a demodataset that is a this pkl file
The idea is the release some code in the future for us to create our own "pkl" to use on EVA3D?
And is there an option to create the texture for the model?
Thanks a lot for the amazing code.
Thank you for your excellent work.
How to simultaneously obtain obj files, corresponding mtl files, and texture images?
Hi,
Thanks so much for this inspiring and excellent work in the Generative Human Model. I have some questions when evaluating the fid of EVA3D. I wonder what is the truncation value to evaluate the method. Is it 1 or the default number of 0.5? If you could provide your script to evaluate all of the metrics of your code, that will be quite helpful for me.
Thanks so much for your help and looking forward to your help.
Best,
Zijian
I want to use my own images for training. May I ask how to estimate the SMLP parameters and camera parameters of an image?How do I get a sample_ Data.pkl?
Thanks for the great work.
Can you please elaborate on how you divide up the volume into boxes?
def predefined_bbox(self, j, only_cur_index=False):
if j == 15:
xyz_min = np.array([-0.0901, 0.2876, -0.0891])
xyz_max = np.array([0.0916, 0.5555+0.04, 0.1390])
xyz_min -= np.array([0.05, 0.05, 0.05])
xyz_max += np.array([0.05, 0.05, 0.05])
cur_index = self.smpl_index_by_joint([15])
elif j == 12:
xyz_min = np.array([-0.1752, 0.0208, -0.1198]) # combine 12 and 9
xyz_max = np.array([0.1724, 0.2876, 0.1391])
cur_index = self.smpl_index_by_joint([9, 13, 14, 6, 16, 17, 12, 15])
elif j == 9 and only_cur_index:
xyz_min = None
xyz_max = None
cur_index = self.smpl_index_by_joint([9, 13, 14, 6, 16, 17, 3])
elif j == 6:
xyz_min = np.array([-0.1569, -0.1144, -0.1095])
xyz_max = np.array([0.1531, 0.0208, 0.1674])
cur_index = self.smpl_index_by_joint([3, 6, 0, 9])
elif j == 3:
xyz_min = np.array([-0.1888, -0.3147, -0.1224])
xyz_max = np.array([0.1852, -0.1144, 0.1679])
cur_index = self.smpl_index_by_joint([3, 0, 1, 2, 6])
elif j == 18:
xyz_min = np.array([0.1724, 0.1450, -0.0750])
xyz_max = np.array([0.4321, 0.2758, 0.0406])
cur_index = self.smpl_index_by_joint([13, 18, 16])
elif j == 20:
xyz_min = np.array([0.4321, 0.1721, -0.0753])
xyz_max = np.array([0.6813, 0.2668, 0.0064])
cur_index = self.smpl_index_by_joint([16, 20, 18])
elif j == 22:
xyz_min = np.array([0.6813, 0.1882, -0.1180])
xyz_max = np.array([0.8731, 0.2445, 0.0461])
cur_index = self.smpl_index_by_joint([22, 20, 18])
elif j == 19:
xyz_min = np.array([-0.4289, 0.1426, -0.0785])
xyz_max = np.array([-0.1752, 0.2754, 0.0460])
cur_index = self.smpl_index_by_joint([14, 17, 19])
elif j == 21:
xyz_min = np.array([-0.6842, 0.1705, -0.0780])
xyz_max = np.array([-0.4289, 0.2659, 0.0059])
cur_index = self.smpl_index_by_joint([17, 19, 21])
elif j == 23:
xyz_min = np.array([-0.8720, 0.1839, -0.1195])
xyz_max = np.array([-0.6842, 0.2420, 0.0465])
cur_index = self.smpl_index_by_joint([23, 21, 19])
elif j == 4:
xyz_min = np.array([0, -0.6899, -0.0849])
xyz_max = np.array([0.1893, -0.3147, 0.1335])
cur_index = self.smpl_index_by_joint([0, 1, 4])
elif j == 7:
xyz_min = np.array([0.0268, -1.0879, -0.0891])
xyz_max = np.array([0.1570, -0.6899, 0.0691])
cur_index = self.smpl_index_by_joint([4, 1, 7])
elif j == 10:
xyz_min = np.array([0.0625, -1.1591-0.04, -0.0876])
xyz_max = np.array([0.1600, -1.0879+0.02, 0.1669])
cur_index = self.smpl_index_by_joint([7, 10, 4])
elif j == 5:
xyz_min = np.array([-0.1935, -0.6964, -0.0883])
xyz_max = np.array([0, -0.3147, 0.1299])
cur_index = self.smpl_index_by_joint([0, 2, 5])
elif j == 8:
xyz_min = np.array([-0.1611, -1.0948, -0.0911])
xyz_max = np.array([-0.0301, -0.6964, 0.0649])
cur_index = self.smpl_index_by_joint([2, 5, 8])
elif j == 11:
xyz_min = np.array([-0.1614, -1.1618-0.04, -0.0882])
xyz_max = np.array([-0.0632, -1.0948+0.02, 0.1680])
cur_index = self.smpl_index_by_joint([8, 11, 5])
else:
xyz_min = xyz_max = cur_index = None
if only_cur_index:
return cur_index
return xyz_min, xyz_max, cur_index
Hi,
Thank you for releasing the great codebase. I am having some issues while running the deepfashion training code. I am keep getting the error "RuntimeError: derivative for aten::grid_sampler_2d_backward is not implemented". My pytorch version is 1.11.0 and cuda version is 11.3. Can you please help regarding this?
I use the released checkpoint models_0420000.pt
and official inference code on DeepFashion dataset.
According to the paper I got 50k inference results. Then I calculate the FlD and KID between results and dataset.
python generation_demo.py --batch 1 --chunk 1 \
--expname 512x256_deepfashion --dataset_path ./dataset/DeepFashion \
--depth 5 --width 128 --style_dim 128 --renderer_spatial_output_dim 512 256 \
--input_ch_views 3 --white_bg \
--voxhuman_name eva3d_deepfashion \
--deltasdf --N_samples 28 --ckpt 420000 \
--identities 50000 #--render_video
For the evaluation code, I use torch-fidelity
package:
fidelity --gpu 0 --kid --fid --input1 ${my_path} --input2 ${dataset_path}
But I only got FID=55, which is much worse than it in the paper.
Am I doing wrong?
Thanks for the great work!
How do you generate the smpl_template_sdf?Is the smpl_template_sdf in part local space and all the part share a same smpl_template_sdf? Could you release the code for generating smpl_template_sdf?
Thanks for great work!
I ran the EVA3D demo code at the google colab, but I could not find RGB 3D meshes.
Can I get RGB 3D meshes or point cloud in this project?
Hi, I think it is a nice work on 3D-Human image generation, so I want to learn more details about the model, but the memory of the graphics card is small so that I try to train the model with AMP (automatic mixed-precision) according to pytorch tutorial, and some problems occurred :
File "/home/xxx/eva3d/op/fused_act.py", line 70, in forward
out = fused.fused_bias_act(input, bias, empty, 3, 0, negative_slope, scale)
RuntimeError: expected scalar type Half but found Float
I think the problem is related to fused_bias_act.cpp but I dont know how to deal with it,couldn‘t appreciate it more if you could give me some advice
hi,
I want to train this model on a similar dataset to deepfashion.
I have calculated keypoint info using openpose(only 18 keypoints), you seems to have more taken keypoint info as input. which method did you use
I am producing segmentation mask using u2net.
in smpl.pkl you have camera parameters and other information, how have you calculated it?
Any other suggestions on how should I proceed?
Excuse me, which code can get the result of novel pose generation(maybe a gif or mp4 :a dance girl). What is the input of the novel pose generation, a dance video of a real person, or a sequence of dance girl's joint position?
Thank you for your outstanding work! May I ask how to determine the length, width, and height of the generated human body in actual space?
I am interested in the processed dataset of DeepFashion. Will you please release it?
I am trying to run EVA3D on collab but the generate code encounters errors while running. The error is missing file or directory in the evaluations folder after downloading the models. Can you please tell me when we start the start collab how to get the software to run? I added a png file in "evaluations/debug/iter_0300000/random_angles/images_paper_fig/PNG_FILE_NAME.png" and it throws FileNotFoundError: No such file or directory.
worth mentioning, I noticed a miss-match between the versions on huggingface and on github.
I am working on thesis on reconstructing 3D models from 2D images and this software is a breakthrough for me, but it requires to look at the code and how it works, so although the huggingface version works I need to work on the collab version that works on github.
your work is amazing, and the help would be greatly appreciated. thanks.
Hi in line1224 in eva3d_deepfashion.py, the shape of cur_input is [1,0,6] many times. That's why
mean some image doesn't have corresponding part? looking forward to your reply
+ python generation_demo.py --batch 1 --chunk 1 --expname 256x256_aist --dataset_path demodataset --depth 5 --width 128 --style_dim 128 --renderer_spatial_output_dim 512 256 --input_ch_views 3 --white_bg --voxhuman_name eva3d_deepfashion --deltasdf --N_samples 28 --ckpt 340000 --identities 5 --truncation_ratio 0.5 --is_aist
Traceback (most recent call last):
File "generation_demo.py", line 18, in <module>
from model import VoxelHumanGenerator as Generator
File "/home/yc/testing3dai/EVA3D/model.py", line 10, in <module>
from op import FusedLeakyReLU, fused_leaky_relu, upfirdn2d
File "/home/yc/testing3dai/EVA3D/op/__init__.py", line 1, in <module>
from .fused_act import FusedLeakyReLU, fused_leaky_relu
File "/home/yc/testing3dai/EVA3D/op/fused_act.py", line 11, in <module>
fused = load(
File "/home/yc/anaconda3/envs/eva3d/lib/python3.8/site-packages/torch/utils/cpp_extension.py", line 986, in load
return _jit_compile(
File "/home/yc/anaconda3/envs/eva3d/lib/python3.8/site-packages/torch/utils/cpp_extension.py", line 1193, in _jit_compile
_write_ninja_file_and_build_library(
File "/home/yc/anaconda3/envs/eva3d/lib/python3.8/site-packages/torch/utils/cpp_extension.py", line 1273, in _write_ninja_file_and_build_library
check_compiler_abi_compatibility(compiler)
File "/home/yc/anaconda3/envs/eva3d/lib/python3.8/site-packages/torch/utils/cpp_extension.py", line 265, in check_compiler_abi_compatibility
if not check_compiler_ok_for_platform(compiler):
File "/home/yc/anaconda3/envs/eva3d/lib/python3.8/site-packages/torch/utils/cpp_extension.py", line 225, in check_compiler_ok_for_platform
which = subprocess.check_output(['which', compiler], stderr=subprocess.STDOUT)
File "/home/yc/anaconda3/envs/eva3d/lib/python3.8/subprocess.py", line 415, in check_output
return run(*popenargs, stdout=PIPE, timeout=timeout, check=True,
File "/home/yc/anaconda3/envs/eva3d/lib/python3.8/subprocess.py", line 516, in run
raise CalledProcessError(retcode, process.args,
subprocess.CalledProcessError: Command '['which', 'c++']' returned non-zero exit status 1.
Hi! Thanks for your great work.
I'm trying to inverse target images as you descibe in 4.5 Inversion section. Could you give me some clue about it?
In detail, I've got an embed with shape(1, 18, 512) at (PTI repo)/embeddings/barcelona/PTI by following https://github.com/danielroich/PTI. However, I just dont know how to change the embed to fit your 'mean_latent' in line 269 generation_demo.py
Hello, I am a newcomer to the world of 3D modeling and NeRF and have a small question. Does the observation space mentioned in the paper refer to the camera space or the space transformed using transformation matrices Gk
(similar to the world space in NeRF)?
Hi,
Thank you very much for your outstanding work. While attempting to train, I noticed that regardless of the number of GPUs selected, the total time required remained constant. If I were to train using a single GPU, should I adjust the iteration count from 1,000,000 to 8,000,000 to account for the fact that you are training with 8 GPUs? Am I correct in understanding that the iteration count pertains to a single GPU?
Very thanks!
I have been trying for a few days and it is not working. I need this software for my thesis topic. Please provide some help.
Thank you for your excellent work.
I searched past issues and read your steps of calculating PCK and Depth metrics. But it seems the evaluation code has not be released yet.
Dear Hong,
Thank you for making your work open source. I am currently attempting to reproduce the results. I trained the model using the recommended command, but even after 1,000,000 iterations, the results remained unsatisfactory. Could you please suggest some possible reasons for this issue?
The results appear as follows:
My command is:
python -m torch.distributed.launch --nproc_per_node ${NUM_GPU} --master_port=${MASTER_PORT} train_deepfashion.py \ --batch 1 --chunk 1 --expname train_deepfashion_512x256_2 --dataset_path ./DeepFashion/ --depth 5 --width 128 --style_dim 128 --renderer_spatial_output_dim 512 256 --input_ch_views 3 --white_bg --r1 300 --voxhuman_name eva3d_deepfashion --random_flip --eikonal_lambda 0.5 --small_aug --iter 1000000 --adjust_gamma --gamma_lb 20 --min_surf_lambda 1.5 --deltasdf --gaussian_weighted_sampler --sampler_std 15 --N_samples 28
Following error when running:
python download_models.py
Downloading EVA3D model pretrained on DeepFashion.
0%| | 2.26k/160M [00:00<5:22:05, 8.30kB/s][Errno Incorrect file size] checkpoint/512x256_deepfashion/volume_renderer/models_0420000.pt
Hi, thanks for your nice work. I notice that you mentioned in the parper you train the model on NVIDIA V100. I want to know that Is it possible to train the model on an NVIDIA 2080ti or 3090?
best wishes
Thanks for your excellent work, fangzhou! Following your paper and other issues, I use 4 V100 and modify training iterations to 2,000,000, and change the batch parameter to 8 in your given train bash file as you mentioned in your paper. But it seems like more than 8000 hours needed to train the model rather than 10 days theoretically. Maybe I should set batch to 1 or just shut down before 1,000,000 iters? It kinda confuses me, please inform me how to adjust my settings.
All the best.
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.