How can machines truly see and understand the three-dimensional world around them? This book takes readers to the frontier of 3D data analysis, offering a compelling exploration of how deep learning transforms raw point clouds into structured, actionable insights across robotics, autonomous driving, architecture, and beyond.

Rather than providing surface-level explanations, this book presents the technical and conceptual foundations of point cloud understanding, from 3D registration and segmentation to object detection and motion tracking. It illuminates how recent advances in neural architectures, feature extraction, and spatial modeling are enabling machines to process unstructured 3D data with increasing precision and efficiency. Readers will discover how these capabilities are reshaping core technologies in navigation, mapping, and intelligent sensing.

Written for researchers, engineers, and graduate students with a background in computer vision, AI, or robotics, the book offers both a rigorous introduction and a deep dive into state-of-the-art solutions. Alongside key methodologies, it addresses open challenges such as noise robustness, cross-domain generalization, and scalabilityinviting readers to engage with the pressing questions driving this fast-evolving field. Whether for academic inquiry or real-world deployment, Point Cloud Intelligence equips professionals with the frameworks and tools needed to lead innovation in intelligent 3D perception.

An overview on point cloud processing, including registration, 3D object detection, segmentation, and generation Suitable for broad audiences, including graduate students, researchers, and professionals in different fields Empower readers to apply advanced technologies to their own research or projects, fostering scientific innovations

Autorentext

Yulan Guo is a full Professor with Sun Yat-sen University. He has authored over 200 articles at highly referred journals and conferences, receiving over 20,000 citations in Google Scholar. His research interests lie in spatial intelligence, 3D vision, and robotics. He served as a Senior Area Editor for IEEE Transactions on Image Processing, and an Associate Editor for the Visual Computer, and Computers & Graphics. He also served as an area chair for CVPR 2025/2023/2021, ICCV 2025/2021, ECCV 2024, NeurIPS 2025/2024, and ACM Multimedia 2021. He organized over 10 workshops, challenges, and tutorials in prestigious conferences such as CVPR, ICCV, ECCV, and 3DV. He is a Senior Member of IEEE and ACM.

Sheng Ao is currently an Assistant Professor with the School of Informatics, Xiamen University, Xiamen, China. He earned his Ph.D. in the School of Electronics and Communication Engineering from the Sun Yat-Sen University (SYSU) in 2024. His research focuses on 3D computer vision, specifically on localization of large-scale 3D point clouds, mapping, and registration. He has contributed to numerous publications in leading journals and conferences such as IEEE TPAMI, IJCV, CVPR, and NeurIPS.

Zhiheng Fu currently is a postdoctoral researcher in the Department of Aeronautical and Aviation Engineering at The HongKong Polytechnic University. He earned his Ph.D. in Computer Science and Software Engineering from the University of Western Australia. He holds a Bachelor of Engineering degree in Electrical Engineering from Northeastern University (NEU) and a Master of Engineering degree in Information and Communication Engineering from the National University of Defense Technology (NUDT). Dr. Fu has published numerous publications in prestigious journals and conferences, including IEEE TIP, PR, ICCV, ECCV, and IJCAI. His current research interests lie in 3D Reconstruction and Generation.

Hao Liu is currently serving as a Young Principal Investigator (Zijiang Young Scholar) at the School of Geospatial Artificial Intelligence, East China Normal University (ECNU), China. Prior to this, he was a Research Fellow at the School of Computer Science and Engineering, Nanyang Technological University (NTU), Singapore. He obtained his B.E. degree from the University of Electronic Science and Technology of China (UESTC) in 2016, followed by an M.E. degree from National University of Defense Technology (NUDT) in 2018. Subsequently, he earned his Ph.D. degree fromSun Yat-Sen University (SYSU) in 2023. His research focuses on 3D deep learning and NeRF, with specific interests in 3D object detection and multi-object tracking.

Klappentext

How can machines truly "see" and understand the three-dimensional world around them? This book takes readers to the frontier of 3D data analysis, offering a compelling exploration of how deep learning transforms raw point clouds into structured, actionable insights across robotics, autonomous driving, architecture, and beyond. Rather than providing surface-level explanations, this book presents the technical and conceptual foundations of point cloud understanding, from 3D registration and segmentation to object detection and motion tracking. It illuminates how recent advances in neural architectures, feature extraction, and spatial modeling are enabling machines to process unstructured 3D data with increasing precision and efficiency. Readers will discover how these capabilities are reshaping core technologies in navigation, mapping, and intelligent sensing. Written for researchers, engineers, and graduate students with a background in computer vision, AI, or robotics, the book offers both a rigorous introduction and a deep dive into state-of-the-art solutions. Alongside key methodologies, it addresses open challenges such as noise robustness, cross-domain generalization, and scalability—inviting readers to engage with the pressing questions driving this fast-evolving field. Whether for academic inquiry or real-world deployment, Point Cloud Intelligence equips professionals with the frameworks and tools needed to lead innovation in intelligent 3D perception.

Inhalt

Introduction.- Local Feature Learning for Point Clouds.- Registration of Point Clouds.- 3D Object Detection in Point Clouds.- Semantic Segmentation of Point Clouds.- Single Object Tracking in Point Cloud Sequences.- Multiple Object Sequences.- Object Completion from Point Clouds.- Semantic Instance Cloud Scenes.- Conclusions and Perspectives.