Data Preprocessing Details

[amogh-tiwari]

Hi!

Thank you for your interesting work, and making the codebase public.

I had a query related to the preprocessing. Can you please point to the script used for preprocessing the data and obtaining the grabnet_.npz files. I am particularly interested in how the BPS was computed. I went through #7 already, but couldn't find the BPS computation information there or in any other script in the repo.

It would be really helpful to have the related script as a reference.

Best Regards,
Amogh

[cutepiggy123]

hi do you encounter this when running test.py
Traceback (most recent call last):
File "grabnet/tests/test.py", line 143, in
grabnet = Trainer(cfg=cfg, inference=True)
File "/home/thanh/GrabNet/./grabnet/train/trainer.py", line 75, in init
self.load_data(cfg, inference)
File "/home/thanh/GrabNet/./grabnet/train/trainer.py", line 157, in load_data
ds_test = LoadData(dataset_dir=cfg.dataset_dir, ds_name=ds_name)
File "/home/thanh/GrabNet/./grabnet/data/dataloader.py", line 40, in init
self.ds = self.np2torch(os.path.join(self.ds_path,'grabnet%s.npz'%ds_name))
File "/home/thanh/GrabNet/./grabnet/data/dataloader.py", line 71, in _np2torch
data = np.load(ds_path, allow_pickle=True)
File "/home/thanh/miniconda3/envs/grab/lib/python3.8/site-packages/numpy/lib/npyio.py", line 405, in load
fid = stack.enter_context(open(os_fspath(file), "rb"))
FileNotFoundError: [Errno 2] No such file or directory: 'GrabNet/grab/test/grabnet_test.npz'

[amogh-tiwari]

hi do you encounter this when running test.py Traceback (most recent call last): File "grabnet/tests/test.py", line 143, in grabnet = Trainer(cfg=cfg, inference=True) File "/home/thanh/GrabNet/./grabnet/train/trainer.py", line 75, in init self.load_data(cfg, inference) File "/home/thanh/GrabNet/./grabnet/train/trainer.py", line 157, in load_data ds_test = LoadData(dataset_dir=cfg.dataset_dir, ds_name=ds_name) File "/home/thanh/GrabNet/./grabnet/data/dataloader.py", line 40, in init self.ds = self.np2torch(os.path.join(self.ds_path,'grabnet%s.npz'%ds_name)) File "/home/thanh/GrabNet/./grabnet/data/dataloader.py", line 71, in _np2torch data = np.load(ds_path, allow_pickle=True) File "/home/thanh/miniconda3/envs/grab/lib/python3.8/site-packages/numpy/lib/npyio.py", line 405, in load fid = stack.enter_context(open(os_fspath(file), "rb")) FileNotFoundError: [Errno 2] No such file or directory: 'GrabNet/grab/test/grabnet_test.npz'

No, I haven't come across this issue. But from the error message, it seems to be something related to the file path. Can you check if the file actually exists at GrabNet/grab/test/grabnet_test.npz?

[amogh-tiwari]

Hi!

Thank you for your interesting work, and making the codebase public.

I had a query related to the preprocessing. Can you please point to the script used for preprocessing the data and obtaining the grabnet_.npz files. I am particularly interested in how the BPS was computed. I went through #7 already, but couldn't find the BPS computation information there or in any other script in the repo.

It would be really helpful to have the related script as a reference.

Best Regards, Amogh

@otaheri: On a related note, can you also tell me what preprocessing is done for objects having lesser number of vertices than the desired number of vertices. For example, I run the grab_preprocessing.py script with n_sample_verts=1024. And I come across a mesh with lesser number of vertices (say 400). How do you preprocess this object? I couldn't figure it out from the code.

Thanks in Advance,
Amogh

[cutepiggy123]

Hi thank you for your reply! Can I ask when you unzip the zip files. do you have the obj_info.npy inside data folder. I am missing that file and can not find that

[amogh-tiwari]

Hi thank you for your reply! Can I ask when you unzip the zip files. do you have the obj_info.npy inside data folder. I am missing that file and can not find that

No. Even I had this problem. There is an obj_info.npy file under the grab folder which I think is the same file, but I am not sure. Regardless, I noticed in the code that while obj_info.npy was being loaded, it was never used. So, I just commented out the line where it was being loaded and ran the code.