Static-dynamic disentanglement

[samiragarwala]

Hi! Thank you for the great work! I had a few questions about the static-disentaglement results reported in Figures 3 and 5 of the paper.

• In Figure 3, are the results reported after training on a single monocular sequence? The static-dynamic disentanglement here seems clean without floaters. I was also wondering if there is egocentric motion in the training sequence or if the camera is stationary?
• In Figure 5, it seems that parked cars are disentangled as dynamic objects. Would I be correct in assuming that the given views were observed at multiple times/views including when the cars are not there?

Thank you!