GitHub - lichiconstant/ObjRecRANSAC: RANSAC variant for 3D object recognition in occluded scenes.

Branches Tags

Name		Name	Last commit message	Last commit date
Latest commit History 2 Commits
FindCUDA		FindCUDA
data		data
src		src
CMakeLists.txt		CMakeLists.txt
README		README
main.cpp		main.cpp
main_seg.cpp		main_seg.cpp
main_test_pose.cpp		main_test_pose.cpp
main_test_seg.cpp		main_test_seg.cpp

Repository files navigation

Greedy ObjRecRANSAC
Author: Chi Li
Date: 05/06/2015

Install:
cd ~
git clone https://github.com/lichiconstant/ObjRecRANSAC.giti
cd ObjRecRANSAC && mkdir build && cd build
cmake .. && make -j8

Run (main_demo.cpp):
./SegRansac --c1 0 --c2 1 --p ../data/ln_joint/ --m 4 -v

Note: c1,c2 are the starting and ending indexes of the sample frames saved in the scene/. The segmented point clouds are saved in result/. '-v' is the flag to visualize the result. You can disable it. '--m' determines the method id:

0: standard ObjRecRANSAC
1: Greedy Batch
2: Greedy One
3: Semantic + Standard ObjRecRANSAC
4: Semantic + Greedy Batch
5: Semantic + Greedy One