PointNet2_Demo details

About

Short Name PointNet2_Demo
Long Name Semantic3D segmentation with Open3D and PointNet2
Website https://github.com/IntelVCL/Open3D-PointNet2-Semantic3D
Description Demo project for Semantic3D (semantic-8) segmentation with Open3D and PointNet2. The purpose of this demo project is to showcase the usage of Open3D in deep learning pipelines. Particularly, we used Open3D for (1) Point cloud data loading, writing, and visualization. Open3D provides efficient implementations of various point cloud manipulation methods (2) Data preprocessing, in particular, voxel-based downsampling (3) Point cloud interpolation, in particular, fast nearest neighbor search for label interpolation. The source code will be released soon at https://github.com/IntelVCL/Open3D-PointNet2-Semantic3D. This project is based on the previous submission http://www.semantic3d.net/view_method_detail.php?method=pointnetpp_sem. We thank the authors for sharing the solution.
Reference https://github.com/IntelVCL/Open3D-PointNet2-Semantic3D, Yixing Lao
Hardware 1 Titian X (Pascal), Intel(R) Core(TM) i7-6950X, 64 GB RAM
Used additional training data 0
Last submission 2018-12-04 07:49:49
Is opensource 1
Number of submissions 1

Result

A_IoU OA [s] IoU 1 IoU 2 IoU 3 IoU 4 IoU 5 IoU 6 IoU 7 IoU 8
Result0.6310.85710000.000.8190.7810.6430.5170.7590.3640.4370.726
Rank34442457323

Full Confusion Matrix

L1' L2' L3' L4' L5' L6' L7' L8'
L1765717874123795890121269243169879816534578121694164
L2576325152605879451666128480598021110841552816130622326
L32266976375458251089513668013815687286792397
L4224392692739721861527700192771141033657474695462950
L59936892115383572370709124117464961984291479963414759962967089
L617328051422599508049147057567662746295650623641983381933
L71378419174232050021402878875433191961114080313193871
L847671987944328112628083358382218414768916218344