aharley
commented
1 year ago I don't have a reference and I don't have a qualitative image for this right now, but I think if you train it without this option, you'll produce the effect mentioned in the paper. It becomes especially clear if you create a video visualization: as a parked car shifts in position relative to the ego-car, its shape will morph in a way that looks similar to orientation change.
mingyangShang
Firstly i think the motivation of this paper is very insightful and the conclusion is solid enough, thanks the authors for your effort. I'm curious about the "randomize reference cam augmentation" mentioned in this paper. I found the first camera and the randomized camera are swapped in code to make sure the 0th cam is randomized referenced camera, but i don't understand what the motivation of this augmentation is and how it's done in detail. Can you explain it or provide the reference paper (if exists). Besides, the paper mentioned "We have observed qualitatively that without this augmentation, the segmented cars have a slight bias forcertain orientations in certain positions; with the augmentation added, this bias disappears.", can you show us the badcase with image to make it more intutive?