The original paper is in English. Non-English content has been machine-translated and may contain typographical errors or mistranslations. ex. Some numerals are expressed as "XNUMX".
Copyrights notice
The original paper is in English. Non-English content has been machine-translated and may contain typographical errors or mistranslations. Copyrights notice
이 논문은 레이블이 있는 3% 및 1% 포인트 클라우드에 대한 새로운 그래프와 GECNN(Edge Convolutional Neural Network)을 사용하여 10D 포인트 클라우드의 약하게 감독된 의미론적 분할을 위한 새로운 방법을 제시합니다. 우리의 일반 프레임워크는 병렬 그래프와 에지 집계 체계를 통해 전역 및 로컬 규모 기능을 모두 인코딩하여 의미론적 분할을 용이하게 합니다. 보다 구체적으로, 포인트 클라우드의 전역 규모 그래프 구조 큐는 그래프 컨벌루션 신경망에 의해 캡처되며, 이는 d-차원 특징 삽입 공간. 우리는 3D 포인트 클라우드의 전역 및 로컬 큐를 모두 융합할 수 있는 동적 에지 기능 집계 컨볼루션 신경망에서 파생된 로컬 규모 기능을 통합합니다. 제안된 GECNN 모델은 부분적으로 레이블이 지정된 지점을 기반으로 불완전하고 부정확하며 자체 감독 및 부드러움 제약 조건으로 구성된 포괄적인 목표를 사용하여 학습됩니다. 제안된 접근 방식은 객관적인 손실에 직접적으로 전역 및 로컬 일관성 제약을 적용합니다. 이는 본질적으로 대규모 포인트 클라우드 공간에서 제한된 주석으로 희박한 3D 포인트 클라우드를 분할하는 문제를 처리합니다. ShapeNet 및 S3DIS 벤치마크에 대한 우리의 실험은 매우 제한된 레이블에도 불구하고 대규모 포인트 클라우드 의미론의 효율적인(20 epoch 이내) 학습을 위해 제안된 접근 방식의 효율성을 보여줍니다.
Zifen HE
Kunming University of Science and Technology
Shouye ZHU
Kunming University of Science and Technology
Ying HUANG
Kunming University of Science and Technology
Yinhui ZHANG
Kunming University of Science and Technology
The copyright of the original papers published on this site belongs to IEICE. Unauthorized use of the original or translated papers is prohibited. See IEICE Provisions on Copyright for details.
부
Zifen HE, Shouye ZHU, Ying HUANG, Yinhui ZHANG, "GECNN for Weakly Supervised Semantic Segmentation of 3D Point Clouds" in IEICE TRANSACTIONS on Information,
vol. E104-D, no. 12, pp. 2237-2243, December 2021, doi: 10.1587/transinf.2021EDP7134.
Abstract: This paper presents a novel method for weakly supervised semantic segmentation of 3D point clouds using a novel graph and edge convolutional neural network (GECNN) towards 1% and 10% point cloud with labels. Our general framework facilitates semantic segmentation by encoding both global and local scale features via a parallel graph and edge aggregation scheme. More specifically, global scale graph structure cues of point clouds are captured by a graph convolutional neural network, which is propagated from pairwise affinity representation over the whole graph established in a d-dimensional feature embedding space. We integrate local scale features derived from a dynamic edge feature aggregation convolutional neural networks that allows us to fusion both global and local cues of 3D point clouds. The proposed GECNN model is trained by using a comprehensive objective which consists of incomplete, inexact, self-supervision and smoothness constraints based on partially labeled points. The proposed approach enforces global and local consistency constraints directly on the objective losses. It inherently handles the challenges of segmenting sparse 3D point clouds with limited annotations in a large scale point cloud space. Our experiments on the ShapeNet and S3DIS benchmarks demonstrate the effectiveness of the proposed approach for efficient (within 20 epochs) learning of large scale point cloud semantics despite very limited labels.
URL: https://global.ieice.org/en_transactions/information/10.1587/transinf.2021EDP7134/_p
부
@ARTICLE{e104-d_12_2237,
author={Zifen HE, Shouye ZHU, Ying HUANG, Yinhui ZHANG, },
journal={IEICE TRANSACTIONS on Information},
title={GECNN for Weakly Supervised Semantic Segmentation of 3D Point Clouds},
year={2021},
volume={E104-D},
number={12},
pages={2237-2243},
abstract={This paper presents a novel method for weakly supervised semantic segmentation of 3D point clouds using a novel graph and edge convolutional neural network (GECNN) towards 1% and 10% point cloud with labels. Our general framework facilitates semantic segmentation by encoding both global and local scale features via a parallel graph and edge aggregation scheme. More specifically, global scale graph structure cues of point clouds are captured by a graph convolutional neural network, which is propagated from pairwise affinity representation over the whole graph established in a d-dimensional feature embedding space. We integrate local scale features derived from a dynamic edge feature aggregation convolutional neural networks that allows us to fusion both global and local cues of 3D point clouds. The proposed GECNN model is trained by using a comprehensive objective which consists of incomplete, inexact, self-supervision and smoothness constraints based on partially labeled points. The proposed approach enforces global and local consistency constraints directly on the objective losses. It inherently handles the challenges of segmenting sparse 3D point clouds with limited annotations in a large scale point cloud space. Our experiments on the ShapeNet and S3DIS benchmarks demonstrate the effectiveness of the proposed approach for efficient (within 20 epochs) learning of large scale point cloud semantics despite very limited labels.},
keywords={},
doi={10.1587/transinf.2021EDP7134},
ISSN={1745-1361},
month={December},}
부
TY - JOUR
TI - GECNN for Weakly Supervised Semantic Segmentation of 3D Point Clouds
T2 - IEICE TRANSACTIONS on Information
SP - 2237
EP - 2243
AU - Zifen HE
AU - Shouye ZHU
AU - Ying HUANG
AU - Yinhui ZHANG
PY - 2021
DO - 10.1587/transinf.2021EDP7134
JO - IEICE TRANSACTIONS on Information
SN - 1745-1361
VL - E104-D
IS - 12
JA - IEICE TRANSACTIONS on Information
Y1 - December 2021
AB - This paper presents a novel method for weakly supervised semantic segmentation of 3D point clouds using a novel graph and edge convolutional neural network (GECNN) towards 1% and 10% point cloud with labels. Our general framework facilitates semantic segmentation by encoding both global and local scale features via a parallel graph and edge aggregation scheme. More specifically, global scale graph structure cues of point clouds are captured by a graph convolutional neural network, which is propagated from pairwise affinity representation over the whole graph established in a d-dimensional feature embedding space. We integrate local scale features derived from a dynamic edge feature aggregation convolutional neural networks that allows us to fusion both global and local cues of 3D point clouds. The proposed GECNN model is trained by using a comprehensive objective which consists of incomplete, inexact, self-supervision and smoothness constraints based on partially labeled points. The proposed approach enforces global and local consistency constraints directly on the objective losses. It inherently handles the challenges of segmenting sparse 3D point clouds with limited annotations in a large scale point cloud space. Our experiments on the ShapeNet and S3DIS benchmarks demonstrate the effectiveness of the proposed approach for efficient (within 20 epochs) learning of large scale point cloud semantics despite very limited labels.
ER -