3D Instance Embedding Learning With a Structure-Aware Loss Function for Point Cloud Segmentation

Liang, Zhidong; Yang, Ming; Li, Hao; Wang, Chunxiang

doi:10.1109/lra.2020.3004802

Cited by 44 publications

(22 citation statements)

References 27 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Region proposal methods 3D-BoNet [64], GSPN (R-PointNet) [65], PanopticFusion [66], LIDARSeg [67], 3DSIS [68], GICN [69] Region proposal free methods SGPN [70], MASC [71], Discriminative embeddings [72], MTML [73], Dynamic Region Growing [74], Clus-terNet [75], PointGroup [76], 3D-BEVIS [77], 3D MPA [78], MT-PNet [79], ASIS [80], JSNet [81]…”

Section: Categories Representative Methodsmentioning

confidence: 99%

Deep Learning Methods Applied to 3D Point Clouds Based Instance Segmentation: A Review

Burume¹,

Du²

2021

Preprint

View full text Add to dashboard Cite

Beyond semantic segmentation,3D instance segmentation(a process to delineate objects of interest and also classifying the objects into a set of categories) is gaining more and more interest among researchers since numerous computer vision applications need accurate segmentation processes(autonomous driving, indoor navigation, and even virtual or augmented reality systems…) This paper gives an overview and a technical comparison of the existing deep learning architectures in handling unstructured Euclidean data for the rapidly developing 3D instance segmentation. First, the authors divide the 3D point clouds based instance segmentation techniques into two major categories which are proposal based methods and proposal free methods. Then, they also introduce and compare the most used datasets with regard to 3D instance segmentation. Furthermore, they compare and analyze these techniques performance (speed, accuracy, response to noise…). Finally, this paper provides a review of the possible future directions of deep learning for 3D sensor-based information and provides insight into the most promising areas for prospective research.

show abstract

Section: Categories Representative Methodsmentioning

confidence: 99%

Deep Learning Methods Applied to 3D Point Clouds Based Instance Segmentation: A Review

Burume¹,

Du²

2021

Preprint

View full text Add to dashboard Cite

show abstract

“…First, the T2 filtering error was relatively low, but it was the highest compared to the other tested methods. This error could be reduced further by designing new loss functions or a more advanced network to understand the semantic relationships among local and global points [97,98]. If the T2 error was lower, then we would expect the RMSE of the DTM to also be lower, accordingly.…”

Section: Contributions and Future Workmentioning

confidence: 99%

A Point-Based Fully Convolutional Neural Network for Airborne LiDAR Ground Point Filtering in Forested Environments

Jin

Zhao

et al. 2020

IEEE J. Sel. Top. Appl. Earth Observations Remote Sensing

View full text Add to dashboard Cite

His research focuses on using deep learning and LiDAR technology to solve phenotyping related challenges. Yanjun Su received a B.E. degree in surveying and mapping engineering from China University of Geosciences, Beijing, China, in 2009, a M.S. degree in geographic information science from the institute of Geographic Sciences and Natural Resources Research, Chinese Academy of Sciences, Beijing, in 2012, and a Ph.D. degree in environmental systems from the University of California at Merced, Merced, CA, USA, in 2017. He is an associate Professor in the Institute of Botany, Chinese Academy of Sciences, Beijing. His research focuses on applying geographic information science and remote sensing to understand the influence of anthropogenic activities and global climate change on terrestrial ecosystems, with a particular emphasis on the terrestrial carbon cycle, terrestrial biodiversity, energy balance and land-use/land-cover change.

show abstract

“…We firstly generate initial embeddings for each point by the submanifold sparse convolutional network. Inspired by the work of [62], we obtain discriminative embeddings for each tree from the LiDAR points based on the structure-aware loss function, which considers both the geometric and the embedding information. In order to achieve refined embeddings, we develop an attention-based graph convolutional neural network that aims to automatically choose and aggregate information from neighbors.…”

Section: Segmentation Of Individual Roadside Trees With Deep Metric Learningmentioning

confidence: 99%

Hierarchical Instance Recognition of Individual Roadside Trees in Environmentally Complex Urban Areas from UAV Laser Scanning Point Clouds

Wang

Jiang

Liu

et al. 2020

IJGI

View full text Add to dashboard Cite

Individual tree segmentation is essential for many applications in city management and urban ecology. Light Detection and Ranging (LiDAR) system acquires accurate point clouds in a fast and environmentally-friendly manner, which enables single tree detection. However, the large number of object categories and occlusion from nearby objects in complex environment pose great challenges in urban tree inventory, resulting in omission or commission errors. Therefore, this paper addresses these challenges and increases the accuracy of individual tree segmentation by proposing an automated method for instance recognition urban roadside trees. The proposed algorithm was implemented of unmanned aerial vehicles laser scanning (UAV-LS) data. First, an improved filtering algorithm was developed to identify ground and non-ground points. Second, we extracted tree-like objects via labeling on non-ground points using a deep learning model with a few smaller modifications. Unlike only concentrating on the global features in previous method, the proposed method revises a pointwise semantic learning network to capture both the global and local information at multiple scales, significantly avoiding the information loss in local neighborhoods and reducing useless convolutional computations. Afterwards, the semantic representation is fed into a graph-structured optimization model, which obtains globally optimal classification results by constructing a weighted indirect graph and solving the optimization problem with graph-cuts. The segmented tree points were extracted and consolidated through a series of operations, and they were finally recognized by combining graph embedding learning with a structure-aware loss function and a supervoxel-based normalized cut segmentation method. Experimental results on two public datasets demonstrated that our framework achieved better performance in terms of classification accuracy and recognition ratio of tree.

show abstract

3D Instance Embedding Learning With a Structure-Aware Loss Function for Point Cloud Segmentation

Cited by 44 publications

References 27 publications

Deep Learning Methods Applied to 3D Point Clouds Based Instance Segmentation: A Review

Deep Learning Methods Applied to 3D Point Clouds Based Instance Segmentation: A Review

A Point-Based Fully Convolutional Neural Network for Airborne LiDAR Ground Point Filtering in Forested Environments

Hierarchical Instance Recognition of Individual Roadside Trees in Environmentally Complex Urban Areas from UAV Laser Scanning Point Clouds

Contact Info

Product

Resources

About