Pub Date : 2023-03-01DOI: 10.14358/pers.22-00121r2
Jingming Xia, Yao Zhou, Ling Tan, Yue Ding
Remote sensing scene images are characterized by intra-class diversity and inter-class similarity. When recognizing remote sensing images, traditional image classification algorithms based on deep learning only extract the global features of scene images, ignoring the important role of local key features in classification, which limits the ability of feature expression and restricts the improvement of classification accuracy. Therefore, this paper presents a multi-channel attention fusion network (MCAFNet). First, three channels are used to extract the features of the image. The channel "spatial attention module" is added after the maximum pooling layer of two channels to get the global and local key features of the image. The other channel uses the original model to extract the deep features of the image. Second, features extracted from different channels are effectively fused by the fusion module. Finally, an adaptive weight loss function is designed to automatically adjust the losses in different types of loss functions. Three challenging data sets, UC Merced Land-Use Dataset (UCM), Aerial Image Dataset (AID), and Northwestern Polytechnic University Dataset (NWPU), are selected for the experiment. Experimental results show that our algorithm can effectively recognize scenes and obtain competitive classification results.
{"title":"MCAFNet: Multi-Channel Attention Fusion Network-Based CNN For Remote Sensing Scene Classification","authors":"Jingming Xia, Yao Zhou, Ling Tan, Yue Ding","doi":"10.14358/pers.22-00121r2","DOIUrl":"https://doi.org/10.14358/pers.22-00121r2","url":null,"abstract":"Remote sensing scene images are characterized by intra-class diversity and inter-class similarity. When recognizing remote sensing images, traditional image classification algorithms based on deep learning only extract the global features of scene images, ignoring the important role\u0000 of local key features in classification, which limits the ability of feature expression and restricts the improvement of classification accuracy. Therefore, this paper presents a multi-channel attention fusion network (MCAFNet). First, three channels are used to extract the features of the\u0000 image. The channel \"spatial attention module\" is added after the maximum pooling layer of two channels to get the global and local key features of the image. The other channel uses the original model to extract the deep features of the image. Second, features extracted from different channels\u0000 are effectively fused by the fusion module. Finally, an adaptive weight loss function is designed to automatically adjust the losses in different types of loss functions. Three challenging data sets, UC Merced Land-Use Dataset (UCM), Aerial Image Dataset (AID), and Northwestern Polytechnic\u0000 University Dataset (NWPU), are selected for the experiment. Experimental results show that our algorithm can effectively recognize scenes and obtain competitive classification results.","PeriodicalId":211256,"journal":{"name":"Photogrammetric Engineering & Remote Sensing","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2023-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125883456","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Book Review – Spatial Analysis for Radar Remote Sensing of Tropical Forests by Gianfranco D. Grandi and Elsa Carla De Grandi","authors":"Konrad E. Kern","doi":"10.14358/pers.89.3.145","DOIUrl":"https://doi.org/10.14358/pers.89.3.145","url":null,"abstract":"","PeriodicalId":211256,"journal":{"name":"Photogrammetric Engineering & Remote Sensing","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2023-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116882576","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-03-01DOI: 10.14358/pers.22-00100r2
Xin Jia, Qing Zhu, X. Ge, Ruifeng Ma, Da Zhang, Tao Liu
As a basic asset of highways, guardrails are essential objects in the digital modeling of highways. Therefore, generating the vectorial 3D trajectory of a guardrail from mobile laser scanning (MLS) point clouds is required for real digital modeling. However, most methods limit straight-line guardrails without considering the continuity and accuracy of the guardrails in turnoff and bend areas; thus, a completed 3D trajectory of a guardrail is not available. We use RANDLA-Net for extracting guardrails as preprocessing of MLS point clouds. We perform a region growth strategy based on linear constraints to obtain correct instantiations and a forward direction. The improved Douglas– Puke algorithm is used to simplify the center points of guardrail, and the 3D trajectory of every guardrail can be vectorized using cubic spline curve fitting. The proposed approach is validated on two 3-km case data sets that can completely instantiate MLS point clouds with remarkable effects. Quantitative evaluations demonstrate that the proposed guardrail instantiation algorithm achieves an overall precision and recall of 98.80% and 97.5%, respectively. The generated 3D trajectory can provide a high-precision design standard for the 3D modeling of the guardrail and has been applied to a long highway scene.
{"title":"Robust Guardrail Instantiation and Trajectory Optimization of Complex Highways Based on Mobile Laser Scanning Point Clouds","authors":"Xin Jia, Qing Zhu, X. Ge, Ruifeng Ma, Da Zhang, Tao Liu","doi":"10.14358/pers.22-00100r2","DOIUrl":"https://doi.org/10.14358/pers.22-00100r2","url":null,"abstract":"As a basic asset of highways, guardrails are essential objects in the digital modeling of highways. Therefore, generating the vectorial 3D trajectory of a guardrail from mobile laser scanning (MLS) point clouds is required for real digital modeling. However, most methods limit straight-line\u0000 guardrails without considering the continuity and accuracy of the guardrails in turnoff and bend areas; thus, a completed 3D trajectory of a guardrail is not available. We use RANDLA-Net for extracting guardrails as preprocessing of MLS point clouds. We perform a region growth strategy based\u0000 on linear constraints to obtain correct instantiations and a forward direction. The improved Douglas– Puke algorithm is used to simplify the center points of guardrail, and the 3D trajectory of every guardrail can be vectorized using cubic spline curve fitting. The proposed approach\u0000 is validated on two 3-km case data sets that can completely instantiate MLS point clouds with remarkable effects. Quantitative evaluations demonstrate that the proposed guardrail instantiation algorithm achieves an overall precision and recall of 98.80% and 97.5%, respectively. The generated\u0000 3D trajectory can provide a high-precision design standard for the 3D modeling of the guardrail and has been applied to a long highway scene.","PeriodicalId":211256,"journal":{"name":"Photogrammetric Engineering & Remote Sensing","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2023-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125388852","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"GIS Tips & Tricks – Simple Customizations Can Have a Large Impact","authors":"Alma M. Karlin","doi":"10.14358/pers.89.3.143","DOIUrl":"https://doi.org/10.14358/pers.89.3.143","url":null,"abstract":"","PeriodicalId":211256,"journal":{"name":"Photogrammetric Engineering & Remote Sensing","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2023-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130792739","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Technology Changes During My 60-year Mapping Career","authors":"D. Maune","doi":"10.14358/pers.89.3.129","DOIUrl":"https://doi.org/10.14358/pers.89.3.129","url":null,"abstract":"","PeriodicalId":211256,"journal":{"name":"Photogrammetric Engineering & Remote Sensing","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2023-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115266658","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Protecting the Places We Love: Conservation Strategies for Entrusted Lands and Parks by Breece Robertson","authors":"M. Ramspott","doi":"10.14358/pers.89.2.69","DOIUrl":"https://doi.org/10.14358/pers.89.2.69","url":null,"abstract":"","PeriodicalId":211256,"journal":{"name":"Photogrammetric Engineering & Remote Sensing","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2023-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128829705","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-02-01DOI: 10.14358/pers.22-00084r2
Batuhan Sariturk, Damla Kumbasar, D. Seker
Building segmentation has numerous application areas such as urban planning and disaster management. In this study, 12 CNN models (U-Net, FPN, and LinkNet using EfficientNet-B5 backbone, U-Net, SegNet, FCN, and six Residual U-Net models) were generated and used for building segmentation. Inria Aerial Image Labeling Data Set was used to train models, and three data sets (Inria Aerial Image Labeling Data Set, Massachusetts Buildings Data Set, and Syedra Archaeological Site Data Set) were used to evaluate trained models. On the Inria test set, Residual-2 U-Net has the highest F1 and Intersection over Union (IoU) scores with 0.824 and 0.722, respectively. On the Syedra test set, LinkNet-EfficientNet-B5 has F1 and IoU scores of 0.336 and 0.246. On the Massachusetts test set, Residual-4 U-Net has F1 and IoU scores of 0.394 and 0.259. It has been observed that, for all sets, at least two of the top three models used residual connections. Therefore, for this study, residual connections are more successful than conventional convolutional layers.
{"title":"Comparative Analysis of Different CNN Models for Building Segmentation from Satellite and UAV Images","authors":"Batuhan Sariturk, Damla Kumbasar, D. Seker","doi":"10.14358/pers.22-00084r2","DOIUrl":"https://doi.org/10.14358/pers.22-00084r2","url":null,"abstract":"Building segmentation has numerous application areas such as urban planning and disaster management. In this study, 12 CNN models (U-Net, FPN, and LinkNet using EfficientNet-B5 backbone, U-Net, SegNet, FCN, and six Residual U-Net models) were generated and used for building segmentation.\u0000 Inria Aerial Image Labeling Data Set was used to train models, and three data sets (Inria Aerial Image Labeling Data Set, Massachusetts Buildings Data Set, and Syedra Archaeological Site Data Set) were used to evaluate trained models. On the Inria test set, Residual-2 U-Net has the highest\u0000 F1 and Intersection over Union (IoU) scores with 0.824 and 0.722, respectively. On the Syedra test set, LinkNet-EfficientNet-B5 has F1 and IoU scores of 0.336 and 0.246. On the Massachusetts test set, Residual-4 U-Net has F1 and IoU scores of 0.394 and 0.259. It has been observed that, for\u0000 all sets, at least two of the top three models used residual connections. Therefore, for this study, residual connections are more successful than conventional convolutional layers.","PeriodicalId":211256,"journal":{"name":"Photogrammetric Engineering & Remote Sensing","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2023-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126286334","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-02-01DOI: 10.14358/pers.22-00101r2
Y. Kaya, H. Şenol, Abdurahman Yasin Yiğit, M. Yakar
It is important to determine car density in parking lots, especially in hospitals, large enterprises, and residential areas, which are used intensively, in terms of executing existing management systems and making precise plans for the future. In this study, cars in parking lots were detected using high-resolution unmanned aerial vehicle (UAV) images with deep learning methods. We tested the performance of the two approaches by determining the number of cars in a parking lot using the You Only Look Once (YOLOv3) and Mask Region–Based Convolutional Neural Networks (Mask R-CNN) approaches as deep learning methods and the deep learning tool of Esri ArcGIS Pro. High-resolution UAV images were processed by photogrammetry and used as input products for the R-CNN and YOLOv3 algorithm. Recall, F1 score, precision ratio/uncertainty accuracy, and average producer accuracy of products automatically extracted with the algorithm were determined as 0.862/0.941, 0.874/0.946, 0.885/0.951, and 0.776/0.897 for R-CNN and YOLOv3, respectively.
特别是在医院、大型企业、住宅等使用密集的停车场，确定停车场的车辆密度，对于执行现有的管理制度和制定精确的未来规划非常重要。在本研究中，使用深度学习方法的高分辨率无人机(UAV)图像检测停车场中的汽车。我们使用You Only Look Once (YOLOv3)和Mask - based Convolutional Neural Networks (Mask R-CNN)方法作为深度学习方法和Esri ArcGIS Pro的深度学习工具，通过确定停车场中的汽车数量来测试这两种方法的性能。采用摄影测量技术对高分辨率无人机图像进行处理，作为R-CNN和YOLOv3算法的输入产品。R-CNN和YOLOv3自动提取产品的召回率为0.862/0.941,F1评分为0.874/0.946，精密度/不确定度准确率为0.885/0.951，生产者平均准确率为0.776/0.897。
{"title":"Car Detection from Very High-Resolution UAV Images Using Deep Learning Algorithms","authors":"Y. Kaya, H. Şenol, Abdurahman Yasin Yiğit, M. Yakar","doi":"10.14358/pers.22-00101r2","DOIUrl":"https://doi.org/10.14358/pers.22-00101r2","url":null,"abstract":"It is important to determine car density in parking lots, especially in hospitals, large enterprises, and residential areas, which are used intensively, in terms of executing existing management systems and making precise plans for the future. In this study, cars in parking lots were\u0000 detected using high-resolution unmanned aerial vehicle (UAV) images with deep learning methods. We tested the performance of the two approaches by determining the number of cars in a parking lot using the You Only Look Once (YOLOv3) and Mask Region–Based Convolutional Neural Networks\u0000 (Mask R-CNN) approaches as deep learning methods and the deep learning tool of Esri ArcGIS Pro. High-resolution UAV images were processed by photogrammetry and used as input products for the R-CNN and YOLOv3 algorithm. Recall, F1 score, precision ratio/uncertainty accuracy, and average producer\u0000 accuracy of products automatically extracted with the algorithm were determined as 0.862/0.941, 0.874/0.946, 0.885/0.951, and 0.776/0.897 for R-CNN and YOLOv3, respectively.","PeriodicalId":211256,"journal":{"name":"Photogrammetric Engineering & Remote Sensing","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2023-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124037053","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-02-01DOI: 10.14358/pers.22-00089r2
R. Sahoo, Shalini Gakhar, R. Rejith, R. Ranjan, M. C. Meena, A. Dey, J. Mukherjee, R. Dhakar, Sunny Arya, Anchal Daas, S. Babu, P. K. Upadhyay, Kapila Sekhawat, Sudhirkumar, Mahesh Kumar, V. Chinnusamy, M. Khanna
Quantitative estimation of crop nitrogen is the key to site-specific management for enhanced nitrogen (N) use efficiency and a sustainable crop production system. As an alternate to the conventional approach through wet chemistry, sensor-based noninvasive, rapid, and near-real-time assessment of crop N at the field scale has been the need for precision agriculture. The present study attempts to predict leaf N of wheat crop through spectroscopy using a field portable spectroradiometer (spectral range of 400–2500 nm) on the ground in the crop field and an imaging spectrometer (spectral range of 400–1000 nm) from an unmanned aerial vehicle (UAV) with the objectives to evaluate (1) four multivariate spectral models (i.e., artificial neural network, extreme learning machine [ELM], least absolute shrinkage and selection operator, and support vector machine regression) and (2) two sets of hyperspectral data collected from two platforms and two different sensors. In the former part of the study, ELM outperforms the other methods with maximum calibration and validation R2 of 0.99 and 0.96, respectively. Furthermore, the image data set acquired from UAV gives higher performance compared to field spectral data. Also, significant bands are identified using stepwise multiple linear regression and used for modeling to generate a wheat leaf N map of the experimental field.
{"title":"Unmanned Aerial Vehicle (UAV)–Based Imaging Spectroscopy for Predicting Wheat Leaf Nitrogen","authors":"R. Sahoo, Shalini Gakhar, R. Rejith, R. Ranjan, M. C. Meena, A. Dey, J. Mukherjee, R. Dhakar, Sunny Arya, Anchal Daas, S. Babu, P. K. Upadhyay, Kapila Sekhawat, Sudhirkumar, Mahesh Kumar, V. Chinnusamy, M. Khanna","doi":"10.14358/pers.22-00089r2","DOIUrl":"https://doi.org/10.14358/pers.22-00089r2","url":null,"abstract":"Quantitative estimation of crop nitrogen is the key to site-specific management for enhanced nitrogen (N) use efficiency and a sustainable crop production system. As an alternate to the conventional approach through wet chemistry, sensor-based noninvasive, rapid, and near-real-time\u0000 assessment of crop N at the field scale has been the need for precision agriculture. The present study attempts to predict leaf N of wheat crop through spectroscopy using a field portable spectroradiometer (spectral range of 400–2500 nm) on the ground in the crop field and an imaging\u0000 spectrometer (spectral range of 400–1000 nm) from an unmanned aerial vehicle (UAV) with the objectives to evaluate (1) four multivariate spectral models (i.e., artificial neural network, extreme learning machine [ELM], least absolute shrinkage and selection operator, and support vector\u0000 machine regression) and (2) two sets of hyperspectral data collected from two platforms and two different sensors. In the former part of the study, ELM outperforms the other methods with maximum calibration and validation R2 of 0.99 and 0.96, respectively. Furthermore, the image data set acquired\u0000 from UAV gives higher performance compared to field spectral data. Also, significant bands are identified using stepwise multiple linear regression and used for modeling to generate a wheat leaf N map of the experimental field.","PeriodicalId":211256,"journal":{"name":"Photogrammetric Engineering & Remote Sensing","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2023-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131982122","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-02-01DOI: 10.14358/pers.22-00087r2
Jordan Bowman, Lexie Yang, O. Thomas, Jerry Kirk, Andrew M. Duncan, D. Hughes, Shannon Meade
Energy infrastructure assessments are needed within 72 hours of natural disasters, and previous data collection methods have proven too slow. We demonstrate a scalable end-to-end solution using a prototype unmanned aerial system that performs on-the-edge detection, classification (i.e., damaged or undamaged), and geo-location of utility poles. The prototype is suitable for disaster response because it requires no local communication infrastructure and is capable of autonomous missions. Collections before, during, and after Hurricane Ida in 2021 were used to test the system. The system delivered an F1 score of 0.65 operating with a 2.7 s/frame processing speed with the YOLOv5 large model and an F1 score of 0.55 with a 0.48 s/frame with the YOLOv5 small model. Geo-location uncertainty in the bottom half of the frame was ∼8 m, mostly driven by error in camera pointing measurement. With additional training data to improve performance and detect additional types of features, a fleet of similar drones could autonomously collect actionable post-disaster data.
{"title":"UAS Edge Computing of Energy Infrastructure Damage Assessment","authors":"Jordan Bowman, Lexie Yang, O. Thomas, Jerry Kirk, Andrew M. Duncan, D. Hughes, Shannon Meade","doi":"10.14358/pers.22-00087r2","DOIUrl":"https://doi.org/10.14358/pers.22-00087r2","url":null,"abstract":"Energy infrastructure assessments are needed within 72 hours of natural disasters, and previous data collection methods have proven too slow. We demonstrate a scalable end-to-end solution using a prototype unmanned aerial system that performs on-the-edge detection, classification (i.e.,\u0000 damaged or undamaged), and geo-location of utility poles. The prototype is suitable for disaster response because it requires no local communication infrastructure and is capable of autonomous missions. Collections before, during, and after Hurricane Ida in 2021 were used to test the system.\u0000 The system delivered an F1 score of 0.65 operating with a 2.7 s/frame processing speed with the YOLOv5 large model and an F1 score of 0.55 with a 0.48 s/frame with the YOLOv5 small model. Geo-location uncertainty in the bottom half of the frame was ∼8 m, mostly driven by error in camera\u0000 pointing measurement. With additional training data to improve performance and detect additional types of features, a fleet of similar drones could autonomously collect actionable post-disaster data.","PeriodicalId":211256,"journal":{"name":"Photogrammetric Engineering & Remote Sensing","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2023-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124939351","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：