/Segbased_Detection

D. Z. Wang, from Segmentation-based Detection to Vote3D

Primary LanguageC++

INPUT DATA

  • use the localization result in the submap of Laser T&R
  • and the submap, and a raw scan of VLP16

PROCESS

  • use the kdtree to remove the background, keep the dynamic ojects(people, cars, etc) on.
  • Cluster on the objects.

Results

  • laserScan on the subMap:

  • Use Knn to get the outliers:

  • the cluster results:

  • Process:

PROBLEM

  • There is problem in the clustering algorithm.
  • Maybe the true EMST is really needed.
  • Many of cluster methods are waiting for me!

Thanks