Tsinghua-MARS-Lab / Occ3D

MIT License
399 stars 25 forks source link

Question about Generating Occupancy Annotation #32

Open yanglong2000 opened 11 months ago

yanglong2000 commented 11 months ago

Hello,

The K-nearest neighbor (KNN) algorithm is used to assign semantic labels to each unlabeled point. In this way, the data density can be improved, but it is bound to produce additional errors, especially the shadow caused by dynamic objects (frame w/o labels), how to solve this problem.

Agito555 commented 6 months ago

My understanding is that KNN is used to assign labels to static objects. For dynamic objects, it may involve interpolating the boxes between two keyframes to obtain the boxes of non-keyframes, then cropping out the corresponding point clouds for non-intermediate frame boxes, and finally assigning labels to the corresponding point clouds. Hope it helps you.