An alternative question I have is, when do we use Object oriented splitting versus grids? I understand each one has tradeoffs, but do they also have certain situations where they perform particularly well? And if so, when? I would assume that splitting based on object groups would do well when there are a lot more objects, but I also feel like that may be slow to compute, so I'm not completely sure.
NKJEW
@aramk-hub I'm guessing that the optimality comes down to how much "overlap" you have between different bounding regions. For grids, if there are lots of objects which span multiple cells, that probably means that it'll be less effective than using the objects themselves. But if the object bounding boxes tend to overlap often, maybe a grid method would be better. I feel like this doesn't really account for all scenarios though, so I might be missing something.
*edit: there are also some student answers on slide 41, it seems
An alternative question I have is, when do we use Object oriented splitting versus grids? I understand each one has tradeoffs, but do they also have certain situations where they perform particularly well? And if so, when? I would assume that splitting based on object groups would do well when there are a lot more objects, but I also feel like that may be slow to compute, so I'm not completely sure.
@aramk-hub I'm guessing that the optimality comes down to how much "overlap" you have between different bounding regions. For grids, if there are lots of objects which span multiple cells, that probably means that it'll be less effective than using the objects themselves. But if the object bounding boxes tend to overlap often, maybe a grid method would be better. I feel like this doesn't really account for all scenarios though, so I might be missing something.
*edit: there are also some student answers on slide 41, it seems