Article
Computer Science, Artificial Intelligence
Zijin Du, Hailiang Ye, Feilong Cao
Summary: This paper introduces a novel convolution block EPFM-Conv, which effectively integrates graph-based method and point-based strategy for extracting local and global features of point cloud. By constructing dynamic graphs and designing edge and point branches, rich detailed features are extracted, while grouped residual learning is used to deepen the network.
KNOWLEDGE-BASED SYSTEMS
(2021)
Article
Environmental Sciences
Shaolei Liu, Kexue Fu, Manning Wang, Zhijian Song
Summary: In this paper, a transformer-based network for point cloud learning is proposed, which effectively models global and local information and enhances the feature representation through radius-based density features. Extensive evaluation demonstrates the effectiveness and competitive performance of the proposed method in point cloud classification and part segmentation.
Article
Engineering, Electrical & Electronic
Tiecheng Sun, Guanghui Liu, Ru Li, Shuaicheng Liu, Shuyuan Zhu, Bing Zeng
Summary: In this paper, a novel point-to-surface representation for 3D point cloud learning is introduced. The method learns a set of quadratic terms to describe 3D shapes and builds connections between local points and global reference surfaces. Experimental results demonstrate the effectiveness of the proposed method in 3D classification and segmentation tasks.
IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS FOR VIDEO TECHNOLOGY
(2022)
Article
Computer Science, Artificial Intelligence
Davide Boscaini, Fabio Poiesi
Summary: Recent trends in deep learning for 3D point cloud understanding focus on proposing sophisticated architectures to capture 3D geometries and introducing inductive biases. However, these architectures often lack evaluation in terms of their ability to generalize to different domains. In this work, we propose PatchMixer, a simple yet effective architecture for 3D point clouds, which processes local patches and uses an MLP for feature aggregation. Our method achieves superior generalization performance on shape classification and part segmentation tasks compared to other deep architectures.
IMAGE AND VISION COMPUTING
(2023)
Article
Robotics
Jintao Chen, Yan Zhang, Feifan Ma, Zhuangbin Tan
Summary: Thanks to the development of deep learning technology and computer science, 3D point cloud analysis has become a research hotspot. This paper proposes an Error feature Back-projection based Local-Global (EB-LG) feature learning module, which captures hidden features and enhances local features. The lightweight and easy-to-use module can be integrated into existing networks to boost their performance, as demonstrated by extensive evaluations on both synthetic and real-world 3D point cloud benchmarks.
IEEE ROBOTICS AND AUTOMATION LETTERS
(2023)
Article
Computer Science, Artificial Intelligence
Shi Qiu, Saeed Anwar, Nick Barnes
Summary: With the help of deep learning, we propose a plug-and-play module, PnP-3D, to improve the effectiveness of existing point cloud networks in analyzing point cloud data. Through experiments on three standard point cloud analysis tasks, we show that PnP-3D can significantly boost the performances of established networks, achieving state-of-the-art results on widely used point cloud benchmarks.
IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE
(2023)
Article
Computer Science, Artificial Intelligence
Tong He, Chunhua Shen, Anton van den Hengel
Summary: In this paper, a simple yet effective approach is proposed for instance segmentation on 3D point cloud. By adopting a top-down strategy and dynamic convolution, instance-aware parameters are generated to improve representation capability. A lightweight transformer is built to capture long-range dependencies. With only non-maximum suppression as post-processing, promising performance is achieved on various datasets.
IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE
(2023)
Article
Environmental Sciences
Genping Zhao, Weiguang Zhang, Yeping Peng, Heng Wu, Zhuowei Wang, Lianglun Cheng
Summary: This paper introduces an efficient deep neural network for 3D LiDAR point cloud classification, achieving effective feature learning and relationship encoding through Point Expanded Grouping and Spatial Embedding units. Compared to other methods, the accuracy improved by 2% while achieving a 26% increase in efficiency on public datasets.
Article
Computer Science, Artificial Intelligence
Jiangjiang Gao, Jinhui Lan, Bingxu Wang, Feifan Li
Summary: This article introduces a novel network model called Spatial Depth Attention Network (SDANet) to improve the accuracy of point cloud classification and segmentation. The model embeds a local depth attention mechanism into the MLP layer to learn local geometric representation of point clouds, and combines it with feature learning to effectively capture the local geometric structure. Experimental results show that SDANet achieves the same or better performance as the state-of-the-art methods on several challenging benchmark datasets and tasks.
Article
Computer Science, Information Systems
Jinlai Zhang, Lyujie Chen, Binbin Liu, Bo Ouyang, Qizhi Xie, Jihong Zhu, Weiming Li, Yanmei Meng
Summary: Recently, it has been discovered that 3D deep learning models are vulnerable to adversarial attacks similar to their 2D counterparts. Existing adversarial attacks on 3D models mostly perform perturbations on 3D point clouds. However, when these attacks are reproduced in physical scenarios, reconstructing the generated adversarial point clouds into meshes significantly reduces their adversarial effects. To address this problem, this paper proposes a strong 3D adversarial attack called Mesh Attack, which directly perturbs the mesh of a 3D object. To maximize the effectiveness of the gradient-based attack, a differentiable sample module that back-propagates the gradient from point cloud to mesh is introduced. Furthermore, three mesh losses are adopted to ensure the generated adversarial mesh examples are free of outliers and 3D printable. Extensive experiments show that the proposed scheme outperforms existing state-of-the-art 3D attacks by a significant margin. SOTA performance is also achieved under various defense mechanisms. The code for this attack is available at: https://github.com/cuge1995/Mesh-Attack.
INFORMATION SCIENCES
(2023)
Article
Environmental Sciences
Yong Feng, Ka Lun Leung, Yingkui Li, Kwai Lam Wong
Summary: This article introduces an AI-based workflow for the registration of large point cloud data. By detecting stable objects from photos and registering only the point cloud data of these objects, the accuracy and computational speed of the registration process are improved.
Article
Computer Science, Artificial Intelligence
Gaihua Wang, Qianyu Zhai, Hong Liu
Summary: This paper introduces a cross self-attention network (CSANet) for raw point cloud classification and segmentation tasks. It has permutation invariance and can learn the coordinates and features of point cloud simultaneously. To capture features at different scales, a multi-scale fusion (MF) module is proposed, which adaptively considers information from different scales and brings richer gradient information, achieving competitive results.
KNOWLEDGE-BASED SYSTEMS
(2022)
Article
Chemistry, Analytical
Qiang Zheng, Jian Sun, Wei Chen
Summary: This paper explores the relationship between local features and their spatial characteristics, and proposes a concise architecture to effectively integrate them in point cloud analysis. The proposed model achieves comparable performance with previous state-of-the-art methods by extracting features and supplementing distribution information.
Review
Computer Science, Hardware & Architecture
Huang Zhang, Changshuo Wang, Shengwei Tian, Baoli Lu, Liping Zhang, Xin Ning, Xiao Bai
Summary: "Point cloud representation" has become a research hotspot in computer vision and has wide applications in fields such as autonomous driving, virtual reality, and robotics. While deep learning techniques have achieved success in processing regular 2D grid image data, they face challenges in handling irregular, unstructured point cloud data. This paper aims to provide researchers with the latest progress and trends in point cloud classification. It introduces point cloud acquisition, characteristics, and challenges, reviews 3D data representations and storage formats, presents deep learning-based methods and recent research work, analyzes the performance of main methods, and discusses challenges and future directions.
Article
Geochemistry & Geophysics
Yupeng Song, Fazhi He, Yansong Duan, Tongzhen Si, Junwei Bai
Summary: This article proposes an enhanced local semantic learning transformer for 3-D point cloud analysis, which can improve the handling of complex point cloud tasks by introducing a local semantic learning point cloud transformer and a local semantic learning self-attention mechanism.
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING
(2022)
Editorial Material
Engineering, Multidisciplinary
Xinyu Li, Pai Zheng, Jinsong Bao, Liang Gao, Xun Xu
Article
Engineering, Industrial
Jiaxin Fan, Chunjiang Zhang, Weiming Shen, Liang Gao
Summary: This paper investigates a flexible job shop scheduling problem with lot-streaming and machine reconfigurations (FJSP-LSMR) for the total weighted tardiness minimisation. A matheuristic method with a variable neighbourhood search component (MH-VNS) is developed to address the problem. The proposed MH-VNS can well balance the solution quality and computational costs for reasonably integrating the GA- and MILP-based local search strategies.
INTERNATIONAL JOURNAL OF PRODUCTION RESEARCH
(2023)
Article
Energy & Fuels
Wei Li, Ningbo Wang, Akhil Garg, Liang Gao
Summary: This paper investigates an air cooling BTMS with 32 cylindrical lithium-ion batteries, focusing on the economic cost caused by parasitic power consumption. By establishing a battery degradation model and using computational fluid dynamics simulation, the study proposes a method to evaluate economy and finds the optimal solution through optimization. The results show a reduction of 0.36K in maximum temperature and 7.6% in cyclical cost.
JOURNAL OF ENERGY STORAGE
(2023)
Article
Automation & Control Systems
Chao Zhao, Weiming Shen
Summary: This article proposes an adversarial mutual information-guided single domain generalization network for machinery fault diagnosis, which learns domain-invariant representations to address domain shift problems. A domain generation module is designed to generate fake target domains with significant distribution discrepancies, and an iterative min-max game of mutual information is implemented to learn generalized features for resisting unknown domain shift. Extensive diagnosis experiments on two mechanical rigs validated the effectiveness of the proposed method.
IEEE TRANSACTIONS ON INDUSTRIAL INFORMATICS
(2023)
Article
Engineering, Industrial
Haobo Qiu, Yingchun Niu, Jie Shang, Liang Gao, Danyang Xu
Summary: This paper proposes an adaptive degradation stage division strategy and a temporal convolutional network (TCN)-based RUL piecewise estimation method. The method includes three steps: extracting features from different domains and selecting highly correlated features with bearing degradation, adaptively dividing the whole lifecycle of bearing into different degradation stages, and establishing a TCN-based piecewise degradation model for accurate prediction of bearing RUL.
JOURNAL OF MANUFACTURING SYSTEMS
(2023)
Article
Thermodynamics
Qixuan Zhong, Parthiv K. Chandra, Wei Li, Liang Gao, Akhil Garg, Song Lv, K. Tai
Summary: This article focuses on the problem of fluctuating cooling system flow caused by different working states during the operation of electric vehicles. The authors propose a two-dimensional topology optimization method for obtaining cooling plates with different topological structures. The results indicate that the optimized cooling plate structure under low flow conditions has better heat dissipation performance.
APPLIED THERMAL ENGINEERING
(2024)
Article
Computer Science, Interdisciplinary Applications
Kaipu Wang, Jun Guo, Baigang Du, Yibing Li, Hongtao Tang, Xinyu Li, Liang Gao
Summary: Collaborative optimization of disassembly line balancing and disassembly sequence planning, combined with a partial destructive disassembly mode, is proposed. A mixed integer linear programming model and a multi-objective improved genetic algorithm are developed. The effectiveness of the proposed model and algorithm is verified in both small-scale and real case studies, and significant improvements are observed in terms of reducing stations, improving smoothness, increasing profits, and reducing energy consumption.
COMPUTERS & INDUSTRIAL ENGINEERING
(2023)
Article
Electrochemistry
Liezhi Lu, Ana Jorge Sobrido, Liang Gao, Akhil Garg, Wei Li
Summary: This paper proposes a physics-based simulation and multi-objective optimization approach for reducing both capacity decay and voltage loss of the Vanadium redox flow battery. The study shows that reducing the electrolyte flow rate and electrode fiber diameter can decrease capacity decay but increase voltage loss. A novel optimization framework is introduced to simultaneously reduce both capacity decay and voltage loss.
ELECTROCHIMICA ACTA
(2023)
Article
Thermodynamics
Xuefei Yang, Hao Li, Liang Gao
Summary: In this paper, a density-based topology optimization method is proposed to optimize the design of multi-phase infill structures by optimizing the pseudo-density and porosity of each element. The method interpolates the thermal stress coefficient to express the relationship between thermal stress load and design variables. The upper bound of solid material volume fraction is constrained to generate sparse but stable structures. Improved weighting method is used to aggregate multiple objective functions. Numerical examples demonstrate the feasibility and effectiveness of the proposed method.
INTERNATIONAL JOURNAL OF HEAT AND MASS TRANSFER
(2023)
Article
Engineering, Biomedical
Liang Gao, Riccardo Beninatto, Tamas Olah, Lars Goebel, Ke Tao, Rebecca Roels, Steffen Schrenker, Julianne Glomm, Jagadeesh K. Venkatesan, Gertrud Schmitt, Ebrar Sahin, Ola Dahhan, Mauro Pavan, Carlo Barbera, Alba Di Lucia, Michael D. Menger, Matthias W. Laschke, Magali Cucchiarini, Devis Galesso, Henning Madry
Summary: This study investigates the short-term safety and efficacy of two novel hyaluronic acid (HA)-triethylene glycol (TEG)-coumarin hydrogels photocrosslinked in situ in a clinically relevant large animal model. It is found that HA hydrogel significantly enhances early cartilage repair and the molar degree of substitution and concentration of HA affects repair.
ADVANCED HEALTHCARE MATERIALS
(2023)
Article
Engineering, Industrial
Han Wang, Min Liu, Weiming Shen
Summary: Manufacturing enterprises are exploring the utilization of industrial knowledge and unlabelled data to achieve human-cyber-physical collaborative and autonomous intelligence. The concept of Industrial-GPT is introduced to solve various tasks in intelligent manufacturing systems. Model as a Service is also introduced to provide a more efficient and flexible service approach. The operation mechanism and challenges of Industrial-GPT in the manufacturing industry are discussed.
IET COLLABORATIVE INTELLIGENT MANUFACTURING
(2023)
Article
Automation & Control Systems
Fei Ming, Wenyin Gong, Ling Wang, Liang Gao
Summary: This article proposes a new constraint-handling technique tailored for decomposition-based many-objective evolutionary algorithms to effectively solve constrained many-objective optimization problems (CMaOPs). The proposed method, called constrained penalty boundary intersection (CPBI), improves the aggregation function by embedding the normalized overall constraint violation to pursue feasibility. The weight of the normalized overall CV is adaptively adjusted based on the feasible ratio of the current population. Experimental results demonstrate the promising performance of CPBI for different problems.
IEEE TRANSACTIONS ON SYSTEMS MAN CYBERNETICS-SYSTEMS
(2023)
Article
Computer Science, Artificial Intelligence
Chao Zhao, Weiming Shen
Summary: This paper proposes a semantic-discriminative augmentation-driven network for imbalanced domain generalization fault diagnosis, which enhances the model's generalization capabilities through synthesizing reliable samples and optimizing representations.
ADVANCED ENGINEERING INFORMATICS
(2024)
Article
Computer Science, Interdisciplinary Applications
Hongjin Wu, Ruoshan Lei, Yibing Peng, Liang Gao
Summary: Machining feature recognition (MFR) is an important step in computer-aided process planning that infers manufacturing semantics from CAD models. Deep learning methods like AAGNet overcome the limitations of traditional rule-based methods by learning from data and preserving geometric and topological information with a novel representation. AAGNet outperforms other state-of-the-art methods in accuracy and complexity, showing potential as a flexible solution for MFR in CAPP.
ROBOTICS AND COMPUTER-INTEGRATED MANUFACTURING
(2024)
Article
Computer Science, Artificial Intelligence
Haizhu Bao, Quanke Pan, Ruben Ruiz, Liang Gao
Summary: This paper investigates the energy-aware scheduling problem in a distributed blocking flow-shop with sequence-dependent setup times. It proposes a cooperative iterated greedy algorithm based on Q-learning (CIG) to minimize makespan and total energy consumption. Experimental results show that CIG outperforms other competitors in terms of improvement percentages.
SWARM AND EVOLUTIONARY COMPUTATION
(2023)
