DatosGreen doesn't match the expected filenames
Closed this issue · 12 comments
Hello
I am trying to recreate the results using DataDanGreen folder
I got the following error when running run_recons.sh
RuntimeError: Folder "smpl" does not exist for sequence "clips". Full path tested: "data/DataDanGreen/clips/clips_smpl".
Uncomment one of the commands and edit the path
Was hoping to get some help resolving this
Thank you
Hi!
You have to edit run_recons.sh and uncomment one of the example commands
Let me know if it works for you
Hey @AndresCasado
Thank you for getting back. I did edit run_recons.sh file
python3 reconstruction_script.py --dir data/DataDanGreen/
# python3 reconstruction_script.py --dir /path/to/dataset/DatosDan/sequences
echo "Uncomment one of the commands and edit the path"
Here is the DataDanGreen directory
[sg7761@log-2 DataDanGreen]$ tree .
.
0037_rect.txt
5;40m0162.png
m
0033_tshirt.npy
g
rt.npy
0mresult_0083_512.png
m
0127_rect.txt
0224_rect.txt
5;40m0263.png
g
38;5;40m0174_tshirt.npy
m0241.png
38;5;40m0289_tshirt.npy
lt_0120_512.png
0298_512.png
0036_rect.txt
5;40m0145.png
lt_0120_512.png
s.npz
3.png
0m
89_tshirt.npy
[0m
lt_0120_512.png
5;40m0057.png
5;40m0152.png
g
38;5;40m0034_tshirt.npy
38;5;40m0103_tshirt.npy
m0170.png
lt_0122_512.png
5;40m0044.png
5;40m0078.png
0142_rect.txt
38;5;40m0035_tshirt.npy
m0102.png
38;5;40m0152_tshirt.npy
lt_0120_512.png
png
8;5;40m0147.png
m
lt_0166_512.png
8;5;40m0036.png
5;40m0117.png
5;40m0216.png
_rect.txt
38;5;40m0034_tshirt.npy
38;5;40m0122_tshirt.npy
m0189.png
38;5;40m0277_tshirt.npy
[0m
[0m
m0101.png_110.pkl
;5;40mbody_frame0225.obj
e0293.obj
[0m
;5;40mbody_frame0174.obj
hed_frame0049.obj
_frame0115.obj
ame0181.obj
;5;40mbody_frame0142.obj
;40mpushed_frame0030.obj
mpushed_frame0096.obj
;5;40mbody_frame0198.obj
;5;40mbody_frame0189.obj
e0060.obj
;5;40mbody_frame0144.obj
e0212.obj
.obj
j
pushed_frame0050.obj
ushed_frame0191.obj
obj
;40m0159.obj
obj
;40m0156.obj
9.obj
;5;40m0155.obj
obj
;40m0155.obj
obj
;40m0155.obj
obj
;40m0154.obj
obj
;40m0159.obj
obj
;40m0153.obj
[0m
Ok, sorry, I thought you hadn't edited it.
The input is in the reconstruction_input.zip. Check the output of tree:
.
├── dan-301
│ ├── dan-301
│ │ ├── 0001.png
│ │ ├── 0001_rect.txt
│ │ ├── 0002.png
│ │ ├── 0002_rect.txt
│ │ ├── 0003.png
│ │ ├── 0003_rect.txt
│ │ ├── ...
│ │ ├── ...
│ │ ├── ...
│ │ ├── ...
│ │ ├── ...
│ ├── dan-301_expose
│ │ ├── 0001.png_109
│ │ │ ├── 0001.png_109_params.npz
│ │ │ ├── 0001.png_109.ply
│ │ │ ├── hd_imgs.png
│ │ │ ├── hd_orig_overlay.png
│ │ │ └── hd_overlay.png
│ │ ├── 0002.png_129
│ │ │ ├── 0002.png_129_params.npz
│ │ │ ├── 0002.png_129.ply
│ │ │ ├── hd_imgs.png
│ │ │ ├── hd_orig_overlay.png
│ │ │ └── hd_overlay.png
│ │ ├── ...
│ │ │ ├── ...
│ │ │ ├── ...
│ │ │ ├── ...
│ │ │ ├── ...
│ │ │ └── ...
│ ├── dan-301_parsing
│ │ ├── 0001.png
│ │ ├── 0001_tshirt.npy
│ │ ├── 0002.png
│ │ ├── 0002_tshirt.npy
│ │ ├── 0003.png
│ │ ├── 0003_tshirt.npy
│ │ ├── ...
│ │ ├── 0260.png
│ │ └── 0260_tshirt.npy
│ ├── dan-301_pifu
│ │ ├── result_0001_512.png
│ │ ├── result_0002_512.png
│ │ ├── ...
│ │ └── result_0260_512.png
│ └── dan-301_smpl
│ ├── 0001.png_109.obj
│ ├── 0001.png_109.pkl
│ ├── 0002.png_129.obj
│ ├── 0002.png_129.pkl
│ ├── ...
│ ├── 0260.png_191.obj
│ └── 0260.png_191.pkl
├── dan-302
│ ├── ...
│ └── ...
├── dan-303
│ ├── ...
│ └── ...
├── ...
│ ├── ...
│ └── ...
Ah I see, let me rerun pointing to the reconstruction_input folder. Ill update here soon.
I seem to have a new error 😅 in case you have faced this before, let me know how to proceed
(pergamo) Singularity> ./run_recons.sh
Tshirt template: ./data/tshirt_4424verts.obj. Exists? True
The md5 is not the same, but ok
All sequences are correct
Traceback (most recent call last): | 0/260 [00:00<?, ?it/s]
File "/scratch/sg7761/deep-learning/pergamo/reconstruction_script.py", line 649, in <module>
main()
File "/scratch/sg7761/deep-learning/pergamo/reconstruction_script.py", line 645, in main
run_sequence_loading_data(processed_sequence_information)
File "/scratch/sg7761/deep-learning/pergamo/reconstruction_script.py", line 427, in run_sequence_loading_data
for info_dict in info_dicts:
File "/scratch/sg7761/deep-learning/pergamo/reconstruction_script.py", line 539, in process_sequence
'smpl_params': smpl_params[frame],
KeyError: 1
@AndresCasado
Hey, since I was not able to figure out the above error, I believe my dataset setup has gone wrong.
I have placed reconstruction_input as in in a folder called DataDanGreen
The command in the run_recon.sh file is
python3 reconstruction_script.py --dir data/DataDanGreen/reconstruction_input/
Let me know if I got anything wrong.
Thanks
On adding some print statements, I see my smpl_params variable is empty
(pergamo) Singularity> ./run_recons.sh
Tshirt template: ./data/tshirt_4424verts.obj. Exists? True
The md5 is not the same, but ok
All sequences are correct
Saving SMPL-converted poses paths: 0%| | 0/520 [00:00<?, ?it/s]
Saving silhouettes paths: 0%| | 0/520 [00:00<?, ?it/s]
hiving Pifu normals paths: 0%| | 0/260 [00:00<?, ?it/s]
smpl_params = {}
<class 'dict'>
hello?
Traceback (most recent call last):
File "/scratch/sg7761/deep-learning/pergamo/reconstruction_script.py", line 654, in <module>
main()
File "/scratch/sg7761/deep-learning/pergamo/reconstruction_script.py", line 650, in main
run_sequence_loading_data(processed_sequence_information)
File "/scratch/sg7761/deep-learning/pergamo/reconstruction_script.py", line 427, in run_sequence_loading_data
for info_dict in info_dicts:
File "/scratch/sg7761/deep-learning/pergamo/reconstruction_script.py", line 544, in process_sequence
'smpl_params': smpl_params[frame],
KeyError: 1
Uncomment one of the commands and edit the path
So on debugging more, I find that the smpl_params expects the pattern of type body_**.pkl with the code here
# region SMPL-converted poses
smpl_params_folder = os.path.join(base_directory, sequence_name, sequence_name + '_smpl')
smpl_pattern = re.compile(r'body_.*?(\d+).*\..*_\d+\.pkl')
smpl_params = {}
for file in tqdm.tqdm(os.listdir(smpl_params_folder), desc="Saving SMPL-converted poses paths"):
match = smpl_pattern.match(file)
if match:
frame = int(match[1])
smpl_path_to_load = os.path.join(smpl_params_folder, file)
smpl_params[frame] = smpl_path_to_load
# endregion
Since none of the files match this pattern, it is always empty.
For reference, I am using this dataset https://urjc-my.sharepoint.com/personal/andres_casado_urjc_es/_layouts/15/onedrive.aspx?ga=1&id=%2Fpersonal%2Fandres%5Fcasado%5Furjc%5Fes%2FDocuments%2FPERGAMO%5Fpublic%2FDataDanGreen
Worked fine when using DatosDanCompressed.zip
Sorry I did not get back earlier. I have checked the folder and, as you stated, it doesn't load the pkl files because they do not match the body_.*.pkl regular expression. I'll keep this issue open because it needs to be fixed, and I'll try to solve it when I can.
Thanks for reporting it!
@AndresCasado Happy to help!!
Hello I am trying to recreate the results using DataDanGreen folder
I got the following error when running
run_recons.shRuntimeError: Folder "smpl" does not exist for sequence "clips". Full path tested: "data/DataDanGreen/clips/clips_smpl". Uncomment one of the commands and edit the pathWas hoping to get some help resolving this
Thank you
@IamShubhamGupto Hello, I apologize for the interruption. I noticed that you have also been working on replicating this code and would like to ask if you have encountered any of the issues I've recently posted about. I look forward to your response.
Hey @Liu-xiaoyu11, unfortunately I have not ran the setup on windows so I don't have enough information