How to recover the shape miss according to the three types (EO, SM, SO) for producing the result of Figure 2 step by step?

Xharlie / BtcDet

Behind the Curtain: Learning Occluded Shapes for 3D Object Detection

Apache License 2.0

195 stars 42 forks source link

How to recover the shape miss according to the three types (EO, SM, SO) for producing the result of Figure 2 step by step? #13

Closed jialeli1 closed 2 years ago

jialeli1 commented 2 years ago

Hi!

How to recover the shape miss according to the three types (EO, SM, SO) for producing the results of Figure 2 step by step?

Many thanks, :)

Xharlie commented 2 years ago

You can follow the source / target shape matching and find closest objects which provide extra points in shape miss regions. Then, you can classify the shape miss regions to 3 categories of EO, SM, SO . These extra points in EO is to recover EO, the extra points in SM is to recover SM.