You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I am trying to fine-tune the LightGlue model with megdaDepth dataset.
But the dataset downloaded from ETH website seems to have some missing and corrupted files:
In scene_info.zip ---> npz files with the following ID cannot be found:
0012
0070
0036
0407
In depth_undistorted.tar.gz ---> Cannot extract complete files from compressed tar.gz file
Only the files in one folder: 0065 can be extracted
Only can extract 500 MB h5 files from a 148 GB compressed tar.gz file
The extraction will stopped after 0065:
Separate two step extraction:
# (1) Extract with gzip first, then untar
$ gunzip -kv depth_undistorted.tar.gz
(after running for a while)
gzip: depth_undistorted_backup.tar.gz: invalid compressed data--format violated
Direct one-step extraction:
# (1) Extract directly with tar
$ tar -xvzf depth_undistorted.tar.gz
depth_undistorted/
depth_undistorted/0065/
depth_undistorted/0065/169836871_03dcb437c5_o.h5
.....
depth_undistorted/0065/2700367171_f26cbce468_o.h5
tar: Skipping to next header
tar: Exiting with failure status due to previous errors
I tried various of extraction methods and none of them works.
I also re-downloaded the dataset files once, for avoiding the corrupted downloading problem from my side.
There is a period that this ETH dataset downloading site is down and cannot be opened, about 2 months ago, maybe something is wrong during that period?
I would appreciate it very much if you could check the data or provide a way for me to deal with this problem!
Thank you very much!
The text was updated successfully, but these errors were encountered:
YuzheHao
changed the title
Corrupted file: [depth_undistorted.tar.gz], unable to extract
Corrupted dataset file: [depth_undistorted.tar.gz], unable to extract
Oct 2, 2024
I am trying to fine-tune the LightGlue model with megdaDepth dataset.
But the dataset downloaded from ETH website seems to have some missing and corrupted files:
In
scene_info.zip
--->npz
files with the following ID cannot be found:In
depth_undistorted.tar.gz
---> Cannot extract complete files from compressedtar.gz
file0065
can be extractedh5
files from a 148 GB compressedtar.gz
fileSeparate two step extraction:
Direct one-step extraction:
# (1) Extract directly with tar $ tar -xvzf depth_undistorted.tar.gz depth_undistorted/ depth_undistorted/0065/ depth_undistorted/0065/169836871_03dcb437c5_o.h5 ..... depth_undistorted/0065/2700367171_f26cbce468_o.h5 tar: Skipping to next header tar: Exiting with failure status due to previous errors
I tried various of extraction methods and none of them works.
I also re-downloaded the dataset files once, for avoiding the corrupted downloading problem from my side.
There is a period that this ETH dataset downloading site is down and cannot be opened, about 2 months ago, maybe something is wrong during that period?
I would appreciate it very much if you could check the data or provide a way for me to deal with this problem!
Thank you very much!
The text was updated successfully, but these errors were encountered: