PIIE-DSA-Net for 3D Semantic Segmentation of Urban Indoor and Outdoor Datasets

Article Instruments & Instrumentation

Fusion of urban 3D point clouds with thermal attributes using MLS data and TIR image sequences

Jingwei Zhu et al.

Summary: The study proposes a method to generate a thermal point cloud by combining Mobile Laser Scanning and Thermal Infrared images for comprehensive evaluation of the energy efficiency and structural details of buildings.

INFRARED PHYSICS & TECHNOLOGY (2021)

添加到收藏夹

Article Environmental Sciences

PointNet plus plus Network Architecture with Individual Point Level and Global Features on Centroid for ALS Point Cloud Classification

Yang Chen et al.

Summary: This study proposed a modified PointNet++ network architecture that concentrates point-level and global features on the centroid point towards local features to facilitate classification. The approach also utilizes a modified Focal Loss function to address the extremely uneven category distribution on ALS point clouds.

REMOTE SENSING (2021)

添加到收藏夹

Article Environmental Sciences

Multi-Scale Attentive Aggregation for LiDAR Point Cloud Segmentation

Xiaoxiao Geng et al.

Summary: The paper introduces a Multi-Scale Attentive Aggregation Network (MSAAN) for semantic segmentation of LiDAR point clouds, which significantly outperformed state-of-the-art methods in experimental results.

REMOTE SENSING (2021)

添加到收藏夹

Article Environmental Sciences

Three-Dimensional Urban Land Cover Classification by Prior-Level Fusion of LiDAR Point Cloud and Optical Imagery

Yanming Chen et al.

Summary: This study proposes a novel prior-level fusion strategy for three-dimensional urban land cover classification using LiDAR point clouds and optical imagery, achieving higher overall accuracy compared to other fusion strategies. The prior-level fusion strategy directly uses semantic information provided by two-dimensional land cover classification, improving the performance of three-dimensional land cover classification.

REMOTE SENSING (2021)

添加到收藏夹

Article Environmental Sciences

Advanced Power Line Diagnostics Using Point Cloud Data-Possible Applications and Limits

Marek Siranec et al.

Summary: This study systematically analyzes the accuracy of sag recalculation affected by input parameter inaccuracies, emphasizing the importance of considering sag deviation during safety assessment processes.

REMOTE SENSING (2021)

添加到收藏夹

Article Computer Science, Information Systems

RGAM: A novel network architecture for 3D point cloud semantic segmentation in indoor scenes

Xue-Tao Chen et al.

Summary: This paper introduces a novel network architecture called RGAM, with four improvements over existing networks, including multi-scale ring grouping learning, neighborhood information fusion, spatial attention module, and channel attention module. Experimental results show that RGAM has stronger recognition ability in 3D point cloud semantic segmentation compared to existing networks.

INFORMATION SCIENCES (2021)

添加到收藏夹

Article Computer Science, Artificial Intelligence

Deep Learning for LiDAR Point Clouds in Autonomous Driving: A Review

Ying Li et al.

Summary: This article reviews the progress of deep learning in processing LiDAR point cloud data, particularly in the field of autonomous driving, including tasks such as segmentation, detection, and classification. Despite several studies focusing on computer vision topics for autonomous vehicles, there is still a research gap in the application of DL in LiDAR point cloud data for autonomous vehicles.

IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS (2021)

添加到收藏夹

Article Environmental Sciences

Self-Attention in Reconstruction Bias U-Net for Semantic Segmentation of Building Rooftops in Optical Remote Sensing Images

Ziyi Chen et al.

Summary: This study introduces a U-Net architecture that combines self-attention and reconstruction-bias modules to enhance semantic segmentation ability, achieving high IoU scores on the WHU and Massachusetts Building datasets.

REMOTE SENSING (2021)

添加到收藏夹

Article Environmental Sciences

MARE: Self-Supervised Multi-Attention REsu-Net for Semantic Segmentation in Remote Sensing

Valerio Marsocci et al.

Summary: The paper proposes a combination of a self-supervised learning algorithm and a semantic segmentation algorithm, focusing on aerial images, and achieves new encouraging results on the ISPRS Vaihingen dataset.

REMOTE SENSING (2021)

添加到收藏夹

Proceedings Paper Computer Science, Artificial Intelligence

Semantic Segmentation for Real Point Cloud Scenes via Bilateral Augmentation and Adaptive Fusion

Shi Qiu et al.

Summary: The study focuses on semantic segmentation of large-scale point cloud data, enhancing local context and multi-resolution features for accurate segmentation.

2021 IEEE/CVF CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION, CVPR 2021 (2021)

添加到收藏夹

Proceedings Paper Computer Science, Artificial Intelligence

Towards Semantic Segmentation of Urban-Scale 3D Point Clouds: A Dataset, Benchmarks and Challenges

Qingyong Hu et al.

Summary: This paper introduces an urban-scale photogrammetric point cloud dataset with nearly three billion richly annotated points, which covers large areas from three UK cities. The performance of state-of-the-art algorithms on this dataset is extensively evaluated, and a comprehensive analysis of the results is provided, identifying key challenges towards urban-scale point cloud understanding.

2021 IEEE/CVF CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION, CVPR 2021 (2021)

添加到收藏夹

Proceedings Paper Computer Science, Artificial Intelligence

CGA-Net: Category Guided Aggregation for Point Cloud Semantic Segmentation

Tao Lu et al.

Summary: The paper proposes the Category Guided Aggregation (CGA) method, which utilizes different aggregation strategies to handle neighbors of the same and different categories, effectively addressing the issue of previous point cloud semantic segmentation networks not fully training on aggregating points of different categories.

2021 IEEE/CVF CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION, CVPR 2021 (2021)

添加到收藏夹

Article Engineering, Electrical & Electronic

GA-NET: Global Attention Network for Point Cloud Semantic Segmentation

Shuang Deng et al.

Summary: In this paper, a global attention network named GA-Net is proposed for point cloud semantic segmentation, which includes a point-independent global attention module and a point-dependent global attention module. A novel random cross attention block is utilized for each point to learn contextual information of all points. Experimental results demonstrate that the proposed method generally outperforms state-of-the-art methods.

IEEE SIGNAL PROCESSING LETTERS (2021)

添加到收藏夹

Article Computer Science, Artificial Intelligence

PRA-Net: Point Relation-Aware Network for 3D Point Cloud Analysis

Silin Cheng et al.

Summary: Learning intra-region contexts and inter-region relations are crucial for enhancing feature representations in point cloud analysis. The proposed framework PRA-Net integrates these two strategies effectively, demonstrating its effectiveness and generalization ability in various 3D benchmarks.

IEEE TRANSACTIONS ON IMAGE PROCESSING (2021)

添加到收藏夹

Article Computer Science, Software Engineering