AC-VRNN
AC-VRNN copied to clipboard
About InD dataset
Hi, thank you for your nice work. I'm using InD dataset for trajectory prediction and I have questions about your works.
In your paper, I saw this statement "We retain only pedestrians’ trajectories and downsample each scene to obtain 20 time steps in 8 s."
Did you only use pedestrian data among the track id of InD? And I don't understand how you did downsampling. Why you use downsampling? I would appreciate it if you could explain this point.