list index out of range while running test.py

[Omar Irfan Khan]

Hi,

Thanks for the great work yall have done! The paper was a great read along with the supplementary material! I wanted to test the code out for myself. Unfortunately, I am running into an error while running test.py

It seems the list index is out of range and as u can see the length of the scores dont match the length of predictions? I am wondering how do i fix this? Should i limit the scores to match the predictions?

I am running this in a conda environment on windows 10.

Edit: Just wanted to add I did download the required weights as per the readme and the dataset

[Omar Irfan Khan]

After looking in the code, it seems this line in test.py is set to:

After looking into BBDataset, I saw that we can change the type to test so it uses the test_set.csv instead of val_set.csv which contains only 3200 images. I changed my code to this in test.py to make it work:

please let me know if this is the correct way to use the code?

[Yyz7mi]

Yes, you should change the type parameter to test the code on the test set. I will correct the code ASAP and thanks a lot for your correction!

[Omar Irfan Khan]

Thanks for your help! I will close this issue!