Bugs during Data Preprocessing

[xupei0610]

These are two bugs in the data preprocessing code that your implementation borrows from Trajectron++ (see StanfordASL/Trajectron-plus-plus#53 and StanfordASL/Trajectron-plus-plus#65).

I tried to reproduce your results with those bugs fixed. The following are the results I got on ETH/UCY datasets. Could you help confirm if these are correct?

    ADE/FDE (reproduced)    ADE/FDE (reported)
eth:   0.55/0.95 ------------> 0.37/0.69
hotel: 0.17/0.28 ------------> 0.12/0.21
univ:  0.25/0.47 ------------> 0.17/0.37
zara1: 0.22/0.44 ------------> 0.13/0.29
zara2: 0.16/0.33 ------------> 0.10/0.21

Could you please provide results on nuScenes with those bugs fixed?

[YongLi000]

hello, I have a problem. When I run "test.py", it show " FileNotFoundError: [Errno 2] No such file or directory: 'data/PIE_dataset/annotations' ". but I do not know how to handle or get the annotations. please help, thanks.