karfly/learnable-triangulation-pytorch

AssertionError : len(self.keypoints_3d_pred) != len(self)

rrmenon10 opened this issue · 3 comments

Hi,

I followed the instructions for downloading the dataset as mentioned here. But even after following these instructions, I end up with an assertion error at this line. I downloaded the validations results from the google drive link mentioned (file size 5.6M for results_val.pkl; train works fine). Is this because of some error in the validation file link?

Thanks!

This is a notorious problem, happens when you have a different number of images in the dataset from what we had. You're not the first one to face it, so I think we're going to refactor this section entirely, hopefully today.

Hi, @rakesh-menon.

Such problem appears, because in our experiments we use full Human3.6M test-set (all frames) and we published precalculated results for all frames. This repository extracts not all, but some of the frames, and that's why you got size mismatch.

In #42 I uploaded new precalculated results which should work with preprocessing pipeline from this repository. Please, update them and tell if it helped you.

@rakesh-menon
It seems like #42 has provided a quick fix by changing that Google drive file (now it is aligned with our Human3.6M preprocessing instructions). Everything should work now, so I decided to leave the code as is.

Thanks for reporting!