Pub Date : 2024-10-09DOI: 10.1016/j.autcon.2024.105818
Kun Yang , Yan Bao , Jiulin Li , Tingli Fan , Chao Tang
To address the increasing issue of cracks in metro shield tunnels, this paper proposes the YOLOv8-GSD model, which integrates DySnakeConv, BiLevelRoutingAttention, and the Gather-and-Distribute Mechanism with the YOLOv8 algorithm. This model is designed for detecting and segmenting cracks in tunnel linings and employs a pixel grouping method to measure crack length and width. Using a real crack dataset from a subway section in Suzhou, China, comparative experiments with YOLOv8x, BlendMask, SOLOv2, and YOLACT demonstrate that YOLOv8-GSD excels in segmentation performance (AP of 82.4 %) and accuracy (IoU of 0.812). The measured crack dimensions show an error within 5 % compared to actual values, confirming the model's effectiveness. These results highlight the potential of YOLOv8-GSD for enhancing the maintenance and safety of metro tunnels.
{"title":"Deep learning-based YOLO for crack segmentation and measurement in metro tunnels","authors":"Kun Yang , Yan Bao , Jiulin Li , Tingli Fan , Chao Tang","doi":"10.1016/j.autcon.2024.105818","DOIUrl":"10.1016/j.autcon.2024.105818","url":null,"abstract":"<div><div>To address the increasing issue of cracks in metro shield tunnels, this paper proposes the YOLOv8-GSD model, which integrates DySnakeConv, BiLevelRoutingAttention, and the Gather-and-Distribute Mechanism with the YOLOv8 algorithm. This model is designed for detecting and segmenting cracks in tunnel linings and employs a pixel grouping method to measure crack length and width. Using a real crack dataset from a subway section in Suzhou, China, comparative experiments with YOLOv8x, BlendMask, SOLOv2, and YOLACT demonstrate that YOLOv8-GSD excels in segmentation performance (AP of 82.4 %) and accuracy (IoU of 0.812). The measured crack dimensions show an error within 5 % compared to actual values, confirming the model's effectiveness. These results highlight the potential of YOLOv8-GSD for enhancing the maintenance and safety of metro tunnels.</div></div>","PeriodicalId":8660,"journal":{"name":"Automation in Construction","volume":"168 ","pages":"Article 105818"},"PeriodicalIF":9.6,"publicationDate":"2024-10-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142418468","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-10-09DOI: 10.1016/j.autcon.2024.105816
Decheng Wu , Xiaoyu Xu , Rui Li , Xuzhao Peng , Xinglong Gong , Chul-Hee Lee , Penggang Pan , Shiyong Jiang
In the field of curtain wall construction, manual installation presents significant safety hazards and suffers from low efficiency, while automated installation is constrained by the limited localization capabilities of curtain wall installation robots. In this paper, an automated installation solution based on machine vision is proposed, and a detailed discussion of several steps involved is provided. To locate the installation area, DANF, a deep learning-based dual-flow aggregation network designed for curtain wall frame segmentation, is proposed. It employs Transformer for global analysis and CNNs for detailed feature extraction to handle curtain wall frame structures. On the dataset constructed in this paper, DANF achieves an IoU of 85.19 % with a parameter count of only 4.24 M, demonstrating higher accuracy compared to other algorithms. Additionally, a pose-solving method based on the semantic segmentation results of the curtain wall frame is designed to adapt to curtain wall installation scenarios.
{"title":"Curtain wall frame segmentation using a dual-flow aggregation network: Application to robot pose estimation","authors":"Decheng Wu , Xiaoyu Xu , Rui Li , Xuzhao Peng , Xinglong Gong , Chul-Hee Lee , Penggang Pan , Shiyong Jiang","doi":"10.1016/j.autcon.2024.105816","DOIUrl":"10.1016/j.autcon.2024.105816","url":null,"abstract":"<div><div>In the field of curtain wall construction, manual installation presents significant safety hazards and suffers from low efficiency, while automated installation is constrained by the limited localization capabilities of curtain wall installation robots. In this paper, an automated installation solution based on machine vision is proposed, and a detailed discussion of several steps involved is provided. To locate the installation area, DANF, a deep learning-based dual-flow aggregation network designed for curtain wall frame segmentation, is proposed. It employs Transformer for global analysis and CNNs for detailed feature extraction to handle curtain wall frame structures. On the dataset constructed in this paper, DANF achieves an IoU of 85.19 % with a parameter count of only 4.24 M, demonstrating higher accuracy compared to other algorithms. Additionally, a pose-solving method based on the semantic segmentation results of the curtain wall frame is designed to adapt to curtain wall installation scenarios.</div></div>","PeriodicalId":8660,"journal":{"name":"Automation in Construction","volume":"168 ","pages":"Article 105816"},"PeriodicalIF":9.6,"publicationDate":"2024-10-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142418469","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-10-08DOI: 10.1016/j.autcon.2024.105810
Gökhan Kazar , Uğur Yiğit , Kenan Evren Boyabatlı
Cost overruns in maintenance projects should be monitored and effectively managed by construction professionals using proactive systems. To establish more effective proactive systems for addressing cost overruns in maintenance projects, this paper presents a topological approach for machine learning-based prediction, integrated into various machine learning models to enhance the feature selection process. Project data from 1807 public schools renovated between 2016 and 2022 was collected to test the proposed mathematical method. The results indicate that the proposed method demonstrates superior performance in 6 out of 7 machine learning algorithms and hybrid models, achieving higher accuracy. This method will enable construction professionals to establish and achieve more efficient proactive systems for managing cost problems in maintenance projects. In addition, this paper will open new doors for developing effective machine-learning models without using optimization methods for other construction issues such as time, quality, or safety.
{"title":"Predicting maintenance cost overruns in public school buildings using a rough topological approach","authors":"Gökhan Kazar , Uğur Yiğit , Kenan Evren Boyabatlı","doi":"10.1016/j.autcon.2024.105810","DOIUrl":"10.1016/j.autcon.2024.105810","url":null,"abstract":"<div><div>Cost overruns in maintenance projects should be monitored and effectively managed by construction professionals using proactive systems. To establish more effective proactive systems for addressing cost overruns in maintenance projects, this paper presents a topological approach for machine learning-based prediction, integrated into various machine learning models to enhance the feature selection process. Project data from 1807 public schools renovated between 2016 and 2022 was collected to test the proposed mathematical method. The results indicate that the proposed method demonstrates superior performance in 6 out of 7 machine learning algorithms and hybrid models, achieving higher accuracy. This method will enable construction professionals to establish and achieve more efficient proactive systems for managing cost problems in maintenance projects. In addition, this paper will open new doors for developing effective machine-learning models without using optimization methods for other construction issues such as time, quality, or safety.</div></div>","PeriodicalId":8660,"journal":{"name":"Automation in Construction","volume":"168 ","pages":"Article 105810"},"PeriodicalIF":9.6,"publicationDate":"2024-10-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142418467","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Targeted to the challenge of indoor scene understanding for intelligent devices, this paper question focuses on enhancing accuracy in semantic information extraction. A framework including a dual hierarchical attention network, transfer learning, interpretability analysis, and modeling module is applied to segment and reconstruct the indoor scene. A high-rise as-built building case is used to verify the method, the results show that: (1) the method achieves a high mIoU of 0.970 in point cloud segmentation and outperforms state-of-the-art methods, both demonstrating strong performance; (2) the method has sound feature extraction and learning ability in term of the interpretive analysis; (3) the method accelerates by 37 % than manual operations, achieving higher accuracy and efficiency. Overall, the method provides an effective solution to segment multi-class objects for indoor scene understanding and can serve as a basis for automated modeling to contribute to an accurate BIM model with great potential for practical application.
{"title":"Dual hierarchical attention-enhanced transfer learning for semantic segmentation of point clouds in building scene understanding","authors":"Limao Zhang , Zeyang Wei , Zhonghua Xiao , Ankang Ji , Beibei Wu","doi":"10.1016/j.autcon.2024.105799","DOIUrl":"10.1016/j.autcon.2024.105799","url":null,"abstract":"<div><div>Targeted to the challenge of indoor scene understanding for intelligent devices, this paper question focuses on enhancing accuracy in semantic information extraction. A framework including a dual hierarchical attention network, transfer learning, interpretability analysis, and modeling module is applied to segment and reconstruct the indoor scene. A high-rise as-built building case is used to verify the method, the results show that: (1) the method achieves a high mIoU of 0.970 in point cloud segmentation and outperforms state-of-the-art methods, both demonstrating strong performance; (2) the method has sound feature extraction and learning ability in term of the interpretive analysis; (3) the method accelerates by 37 % than manual operations, achieving higher accuracy and efficiency. Overall, the method provides an effective solution to segment multi-class objects for indoor scene understanding and can serve as a basis for automated modeling to contribute to an accurate BIM model with great potential for practical application.</div></div>","PeriodicalId":8660,"journal":{"name":"Automation in Construction","volume":"168 ","pages":"Article 105799"},"PeriodicalIF":9.6,"publicationDate":"2024-10-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142418465","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-10-07DOI: 10.1016/j.autcon.2024.105821
Siwei Lin , Liping Duan , Jiming Liu , Xiao Xiao , Ji Miao , Jincheng Zhao
{"title":"Corrigendum to “Automated geometric reconstruction and cable force inference for cable-net structures using 3D point clouds” [Automation in Construction, 165 (2024), 105543]","authors":"Siwei Lin , Liping Duan , Jiming Liu , Xiao Xiao , Ji Miao , Jincheng Zhao","doi":"10.1016/j.autcon.2024.105821","DOIUrl":"10.1016/j.autcon.2024.105821","url":null,"abstract":"","PeriodicalId":8660,"journal":{"name":"Automation in Construction","volume":"168 ","pages":"Article 105821"},"PeriodicalIF":9.6,"publicationDate":"2024-10-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142586568","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-10-05DOI: 10.1016/j.autcon.2024.105811
Vincent Gbouna Zakka , Minhyun Lee , Ruixiaoxiao Zhang , Lijie Huang , Seunghoon Jung , Taehoon Hong
An efficient method for predicting occupants' thermal comfort is crucial for developing optimal environmental control strategies while minimizing energy consumption in buildings. This paper presents a non-invasive vision-based personal comfort model that integrates thermographic images and deep learning. Unlike previous studies, the entire thermographic image of the upper body is directly used during model training, minimizing complex data processing and maximizing the use of rich skin temperature distribution. The proposed method is validated using thermographic images and corresponding thermal sensation votes (TSV) from 10 participants under different experimental conditions. Results show that the model based on a 3-point TSV scale achieves exceptional classification performance with an average accuracy of 99.51 %, outperforming existing models. The model performance using a 7-point TSV scale is slightly lower, with an average accuracy of 89.90 %. This method offers potential for integrating thermal comfort models into real-time building environmental control, optimizing occupant comfort and energy consumption.
{"title":"Non-invasive vision-based personal comfort model using thermographic images and deep learning","authors":"Vincent Gbouna Zakka , Minhyun Lee , Ruixiaoxiao Zhang , Lijie Huang , Seunghoon Jung , Taehoon Hong","doi":"10.1016/j.autcon.2024.105811","DOIUrl":"10.1016/j.autcon.2024.105811","url":null,"abstract":"<div><div>An efficient method for predicting occupants' thermal comfort is crucial for developing optimal environmental control strategies while minimizing energy consumption in buildings. This paper presents a non-invasive vision-based personal comfort model that integrates thermographic images and deep learning. Unlike previous studies, the entire thermographic image of the upper body is directly used during model training, minimizing complex data processing and maximizing the use of rich skin temperature distribution. The proposed method is validated using thermographic images and corresponding thermal sensation votes (TSV) from 10 participants under different experimental conditions. Results show that the model based on a 3-point TSV scale achieves exceptional classification performance with an average accuracy of 99.51 %, outperforming existing models. The model performance using a 7-point TSV scale is slightly lower, with an average accuracy of 89.90 %. This method offers potential for integrating thermal comfort models into real-time building environmental control, optimizing occupant comfort and energy consumption.</div></div>","PeriodicalId":8660,"journal":{"name":"Automation in Construction","volume":"168 ","pages":"Article 105811"},"PeriodicalIF":9.6,"publicationDate":"2024-10-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142418464","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-10-04DOI: 10.1016/j.autcon.2024.105804
Daniel Eriksson , Reza Ghabcheloo , Marcus Geimer
Reinforcement Learning (RL) requires many interactions with the environment to converge to an optimal strategy, which makes it unfeasible to apply to wheel loaders and the bucket filling problem without using simulators. However, it is difficult to model the pile dynamics in the simulator because of unknown parameters, which results in poor transferability from the simulation to the real environment. Instead, this paper uses world models, serving as a fast surrogate simulator, creating a dream environment where a reinforcement learning (RL) agent explores and optimizes its bucket-filling behavior. The trained agent is then deployed on a full-size wheel loader without modifications, demonstrating its ability to outperform the previous benchmark controller, which was synthesized using imitation learning. Additionally, the same performance was observed as that of a controller pre-trained with imitation learning and optimized on the test pile using RL.
{"title":"Optimizing bucket-filling strategies for wheel loaders inside a dream environment","authors":"Daniel Eriksson , Reza Ghabcheloo , Marcus Geimer","doi":"10.1016/j.autcon.2024.105804","DOIUrl":"10.1016/j.autcon.2024.105804","url":null,"abstract":"<div><div>Reinforcement Learning (RL) requires many interactions with the environment to converge to an optimal strategy, which makes it unfeasible to apply to wheel loaders and the bucket filling problem without using simulators. However, it is difficult to model the pile dynamics in the simulator because of unknown parameters, which results in poor transferability from the simulation to the real environment. Instead, this paper uses world models, serving as a fast surrogate simulator, creating a dream environment where a reinforcement learning (RL) agent explores and optimizes its bucket-filling behavior. The trained agent is then deployed on a full-size wheel loader without modifications, demonstrating its ability to outperform the previous benchmark controller, which was synthesized using imitation learning. Additionally, the same performance was observed as that of a controller pre-trained with imitation learning and optimized on the test pile using RL.</div></div>","PeriodicalId":8660,"journal":{"name":"Automation in Construction","volume":"168 ","pages":"Article 105804"},"PeriodicalIF":9.6,"publicationDate":"2024-10-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142418462","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-10-04DOI: 10.1016/j.autcon.2024.105805
Yoojun Kim , Youngjib Ham
To effectively mitigate heat risks, it is crucial to pinpoint areas of high vulnerability and assess the severity of heat-related threats to construction workers. This paper advances the understanding of heat risks in construction by mapping the associated risks across time and space to support informed decision-making. This paper presents a framework for heat risk monitoring, enabled by a construction site digital twin. This framework leverages geometric modeling, incorporates real-time weather data from a weather station, and utilizes computational simulations for assessing spatio-temporal heat risks. Its effectiveness was validated through a case study in Stephenville, Texas, USA, where it demonstrated superior fidelity when compared to using the conventional black-globe thermometer. Moreover, the results substantiated that incorporating the spatio-temporal variability of heat risks enhances heat risk surveillance in construction workplaces. This approach offers practical insights into imminent heat-related threats, aiming to prevent potential heat-related accidents in construction.
{"title":"Spatio-temporal heat risk analysis in construction: Digital twin-enabled monitoring","authors":"Yoojun Kim , Youngjib Ham","doi":"10.1016/j.autcon.2024.105805","DOIUrl":"10.1016/j.autcon.2024.105805","url":null,"abstract":"<div><div>To effectively mitigate heat risks, it is crucial to pinpoint areas of high vulnerability and assess the severity of heat-related threats to construction workers. This paper advances the understanding of heat risks in construction by mapping the associated risks across time and space to support informed decision-making. This paper presents a framework for heat risk monitoring, enabled by a construction site digital twin. This framework leverages geometric modeling, incorporates real-time weather data from a weather station, and utilizes computational simulations for assessing spatio-temporal heat risks. Its effectiveness was validated through a case study in Stephenville, Texas, USA, where it demonstrated superior fidelity when compared to using the conventional black-globe thermometer. Moreover, the results substantiated that incorporating the spatio-temporal variability of heat risks enhances heat risk surveillance in construction workplaces. This approach offers practical insights into imminent heat-related threats, aiming to prevent potential heat-related accidents in construction.</div></div>","PeriodicalId":8660,"journal":{"name":"Automation in Construction","volume":"168 ","pages":"Article 105805"},"PeriodicalIF":9.6,"publicationDate":"2024-10-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142418460","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-10-04DOI: 10.1016/j.autcon.2024.105802
Yishu Yang , Ying Yu , Chenglin Yu , Ray Y. Zhong
Prefabricated construction is increasingly replacing traditional methods due to its higher productivity, superior quality, and shorter construction time. This paper aims to optimize production and logistics collaboration within a three-tier prefabricated supply chain network to reduce overall costs and enhance response efficiency. A decision model was developed that integrates factory and logistics capacity, on-site assembly sequence, and outsourcing decisions to optimize resource allocation. The model demonstrates superior cost efficiency and resource allocation effectiveness over the Earliest Due Date (EDD) method through a hypothetical case study. This result provides robust decision support for supply chain professionals, offering significant practical implications for cost reduction and resource optimization. Our findings lay a foundation for future studies on supply chain management and optimization under dynamic conditions, offering new perspectives and methodologies.
{"title":"Data-driven logistics collaboration for prefabricated supply chain with multiple factories","authors":"Yishu Yang , Ying Yu , Chenglin Yu , Ray Y. Zhong","doi":"10.1016/j.autcon.2024.105802","DOIUrl":"10.1016/j.autcon.2024.105802","url":null,"abstract":"<div><div>Prefabricated construction is increasingly replacing traditional methods due to its higher productivity, superior quality, and shorter construction time. This paper aims to optimize production and logistics collaboration within a three-tier prefabricated supply chain network to reduce overall costs and enhance response efficiency. A decision model was developed that integrates factory and logistics capacity, on-site assembly sequence, and outsourcing decisions to optimize resource allocation. The model demonstrates superior cost efficiency and resource allocation effectiveness over the Earliest Due Date (EDD) method through a hypothetical case study. This result provides robust decision support for supply chain professionals, offering significant practical implications for cost reduction and resource optimization. Our findings lay a foundation for future studies on supply chain management and optimization under dynamic conditions, offering new perspectives and methodologies.</div></div>","PeriodicalId":8660,"journal":{"name":"Automation in Construction","volume":"168 ","pages":"Article 105802"},"PeriodicalIF":9.6,"publicationDate":"2024-10-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142418461","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-10-04DOI: 10.1016/j.autcon.2024.105806
Luping Li , Jian Chen , Xing Su , Haoying Han , Chao Fan
Semantic segmentation is crucial for interpreting point cloud data and plays a fundamental role in automating the creation of as-built BIM. Existing neural network models for semantic segmentation often heavily rely on the training dataset, resulting in a significant performance drop when applied to new datasets. This paper presents AttTransNet, a neural network model for automated point cloud semantic segmentation. Its attention-based pooling module improves local feature extraction from point clouds while reducing computational costs. The transfer learning framework enhances segmentation accuracy with minimal training on target datasets. Comparative experiments show that AttTransNet reduces training time by 80 % and improves segmentation accuracy by over 20 % compared with other SOTA methods. Cross-dataset experiments reveal that the transfer learning framework increases accuracy on new datasets by 150 %. By adding semantic information to point clouds, AttTransNet aids BIM modelers with direct reference, encouraging broader application of automated point cloud segmentation in the industry.
{"title":"Deep learning network for indoor point cloud semantic segmentation with transferability","authors":"Luping Li , Jian Chen , Xing Su , Haoying Han , Chao Fan","doi":"10.1016/j.autcon.2024.105806","DOIUrl":"10.1016/j.autcon.2024.105806","url":null,"abstract":"<div><div>Semantic segmentation is crucial for interpreting point cloud data and plays a fundamental role in automating the creation of as-built BIM. Existing neural network models for semantic segmentation often heavily rely on the training dataset, resulting in a significant performance drop when applied to new datasets. This paper presents AttTransNet, a neural network model for automated point cloud semantic segmentation. Its attention-based pooling module improves local feature extraction from point clouds while reducing computational costs. The transfer learning framework enhances segmentation accuracy with minimal training on target datasets. Comparative experiments show that AttTransNet reduces training time by 80 % and improves segmentation accuracy by over 20 % compared with other SOTA methods. Cross-dataset experiments reveal that the transfer learning framework increases accuracy on new datasets by 150 %. By adding semantic information to point clouds, AttTransNet aids BIM modelers with direct reference, encouraging broader application of automated point cloud segmentation in the industry.</div></div>","PeriodicalId":8660,"journal":{"name":"Automation in Construction","volume":"168 ","pages":"Article 105806"},"PeriodicalIF":9.6,"publicationDate":"2024-10-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142418463","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
扫码关注我们
求助内容：
应助结果提醒方式：
京公网安备 11010802042870号