hysts / pytorch_mpiigaze

An unofficial PyTorch implementation of MPIIGaze and MPIIFaceGaze
MIT License
346 stars 85 forks source link

Issues in preprocess_mpiigaze.py --> ValueError: Unknown mat file type, version 0, 0 #28

Closed err-or closed 3 years ago

err-or commented 3 years ago

Hi, I am trying to use your preprocessing code for the mpiigaze dataset but I am encountering this issue while running the preprocess_mpiigaze.py file. image

Here, under 'MPIIGaze/' I kept the total mpiigaze data that we can get from here.

FYI, I am using a virtual environment and the scipy version is: 1.6.0

I would appreciate any kind of help to resolve this issue. Thanks!

hysts commented 3 years ago

Hi, @err-or

Hmm, I have no idea what's going on. I downloaded the dataset again and ran the code, and it works fine with scipy 1.6.0 on my Ubuntu environment. It may be an error that occurs in the Windows environment, but as I don't have a Windows environment, I can't check it myself.

I think it's unlikely, but maybe the mat file could be corrupted. Is the md5sum of MPIIGaze/Data/Normalized/p00/day01.mat the same as 7f06339fc5e4e3f0ba8ff76c6d886bfe? And is it not possible to open the file with scipy.io.loadmat even if the md5sum is the same?

err-or commented 3 years ago

Hi @hysts,

Thank you for your prompt response. Yes I have checked the md5sum and they are the same as you mentioned and I can also load the .mat file with scipy.io.loadmat.

The problem was solely on my end as I copied the script out of the repository and running it separately from a different location. My apologies for the inconvenience. Thank you for your help. I am closing the issue.