Submission: PPF_3D_ICP/LM-O/PPF_3D_ICP

Download submission
Submission name PPF_3D_ICP
Submission time (UTC) March 3, 2022, 1:37 p.m.
User yuecideng
Task Model-based 6D localization of seen objects
Dataset LM-O
Training model type CAD
Training image type None
Description use segmentation result to obtain mask of object instance, then use PPF+ICP for pose estimation,
Evaluation scores
AR:0.636
AR_MSPD:0.716
AR_MSSD:0.669
AR_VSD:0.523
average_time_per_image:0.226

Method: PPF_3D_ICP

User yuecideng
Publication
Implementation https://github.com/yuecideng/Misc3D
Training image modalities D
Test image modalities D
Description

A modified implementation of paper "Drost et al., Model globally, match locally: Efficient and robust 3D object recognition, CVPR 2010". Add sparse icp and dense icp for post processing.

Computer specifications intel@core i7-1165G7 2.8Gz 8 threads