Article
Geochemistry & Geophysics
Yuhan Wang, Lingjia Gu, Xiaofeng Li, Ruizhi Ren
Summary: Inspired by recent developments in deep learning networks, this letter presents a novel approach based on a multifeature LSTM network for extracting buildings from multitemporal high-resolution optical satellite imagery. Experimental results show that the proposed method outperforms current deep learning methods for building extraction.
IEEE GEOSCIENCE AND REMOTE SENSING LETTERS
(2021)
Article
Engineering, Electrical & Electronic
Jie Wan, Zhong Xie, Yongyang Xu, Siqiong Chen, Qinjun Qiu
Summary: The DA-RoadNet is a road extraction network with dual attention mechanism, designed to effectively solve discontinuous problems and preserve the integrity of extracted roads by utilizing a deep learning network model and a novel attention mechanism module. Additionally, a hybrid loss function is employed to address class imbalance, ensuring stable training of the network model and avoiding local optima.
IEEE JOURNAL OF SELECTED TOPICS IN APPLIED EARTH OBSERVATIONS AND REMOTE SENSING
(2021)
Article
Geochemistry & Geophysics
Haifeng Li, Yi Li, Guo Zhang, Ruoyun Liu, Haozhe Huang, Qing Zhu, Chao Tao
Summary: This study proposes a global style and local matching contrastive learning network (GLCNet) for remote sensing image (RSI) semantic segmentation. By using global style contrastive learning and local feature matching contrastive learning modules, the method achieves superior results compared to state-of-the-art methods and supervised learning methods on various RSI semantic segmentation datasets.
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING
(2022)
Article
Geochemistry & Geophysics
Yong Cao, Chunlei Huo, Nuo Xu, Xin Zhang, Shiming Xiang, Chunhong Pan
Summary: Semantic segmentation is crucial for understanding very high resolution (VHR) images. This paper proposes a head-level ensemble network (HENet) to address the challenge of integrating deep models. HENet reduces model complexity by sharing feature extraction networks and improves complementarity between models through cooperative learning (CL).
IEEE GEOSCIENCE AND REMOTE SENSING LETTERS
(2022)
Article
Engineering, Electrical & Electronic
Miguel-Angel Manso-Callejo, Calimanut-Ionut Cira, Ramon Pablo Alcarria Garrido, Francisco Javier Gonzalez Matesanz
Summary: Deep learning applied to feature extraction and mapping from high-resolution images shows potential in improving terrain mapping processes. Experiences have been applied on a small scale with high expectations for country-wide applicability. The methodology can also be adapted for large-scale efficient extraction of geospatial elements with automated procedures.
IEEE JOURNAL OF SELECTED TOPICS IN APPLIED EARTH OBSERVATIONS AND REMOTE SENSING
(2021)
Article
Engineering, Electrical & Electronic
Yun-Cheng Li, Heng-Chao Li, Wen-Shuai Hu, Hui-Ling Yu
Summary: The study introduces a novel dual-channel scale-aware segmentation network for effective semantic segmentation of high-resolution aerial images. By utilizing Xception branch and DSMPCF branch to process different types of image information, and incorporating position and channel attention in the proposed model, it achieves more accurate and efficient results.
IEEE JOURNAL OF SELECTED TOPICS IN APPLIED EARTH OBSERVATIONS AND REMOTE SENSING
(2021)
Article
Geochemistry & Geophysics
Boyu Zhao, Mengmeng Zhang, Jianbu Wang, Xiukai Song, Yuanyuan Gui, Yuxiang Zhang, Wei Li
Summary: The study proposes a multiple attention network (MARNet) based on transfer learning to address the segmentation issues of Spartina alterniflora in remote sensing images on wetlands. The method incorporates a plug-and-play attention module to enhance the learning of vegetation features and improve the network's ability to focus on small areas of S. alterniflora. MARNet also designs a transfer learning architecture from both interdomain alignment and intradomain adaptation perspectives to align the statistical distribution and enhance high confidence prediction. Experimental results demonstrate that MARNet outperforms other networks in accurately extracting S. alterniflora in wetlands.
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING
(2023)
Article
Geochemistry & Geophysics
Ming Lu, Leyuan Fang, Muxing Li, Bob Zhang, Yi Zhang, Pedram Ghamisi
Summary: The article explores using point labels and a neighbor feature aggregation network (NFANet) to extract water bodies. Compared to pixel-level labels, point labels are easier to obtain but lose information, thus using neighboring features to extract more representative features.
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING
(2022)
Article
Computer Science, Information Systems
Xungen Li, Zhan Zhang, Shuaishuai Lv, Mian Pan, Qi Ma, Haibin Yu
Summary: This article proposes a road extraction method based on multi-task key point constraints, which improves the accuracy and effectiveness of road extraction by addressing issues such as land coverage, building coverage, and tree shading in remote sensing images. The method utilizes position and channel attention mechanisms, as well as a multi-branch cascade dilated spatial pyramid, to enhance semantic information fusion and solve the problem of information loss during extraction. Experimental results demonstrate that the proposed method outperforms several state-of-the-art techniques in terms of detection accuracy, recall, precision, and F1-score.
Article
Geochemistry & Geophysics
Rui Zhao, Zhenwei Shi, Zhengxia Zou
Summary: An attention-based method that generates pixelwise semantic content segmentation masks for remote sensing images is proposed, achieving higher captioning accuracy compared to state-of-the-art methods. This method utilizes fine-grained, structured attention to exploit the structural characteristics of semantic contents in high-resolution remote sensing images.
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING
(2022)
Article
Geochemistry & Geophysics
Jun Chen, Yuxuan Jiang, Linbo Luo, Wenping Gong
Summary: Building footprint extraction is important in various remote-sensing applications. This article proposes a novel adaptive screening feature network (ASF-Net) to accurately extract building footprints from aerial images. By adjusting the receptive field and enhancing feature information, ASF-Net achieves competitive results in multiscale building footprint inference.
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING
(2022)
Article
Computer Science, Information Systems
Leida Li, Yixuan Li, Jinjian Wu, Lin Ma, Yuming Fang
Summary: This paper proposes a quality evaluation model for image retargeting based on instance semantics, which extracts instance-level semantic features and integrates semantic categories and global features to achieve more accurate quality scores.
IEEE TRANSACTIONS ON MULTIMEDIA
(2021)
Article
Engineering, Electrical & Electronic
Fei Pan, Zebin Wu, Qian Liu, Yang Xu, Zhihui Wei
Summary: A densely connected feature fusion network (DCFF-Net) is proposed for high-resolution remote sensing image change detection. It extracts multiscale raw image features and difference information using a two-stream network, employs attention mechanism and deep supervision strategy in reconstructing change maps, and introduces a novel weighted loss to alleviate data imbalance issue. Extensive experiments confirm the superiority of the proposed method over other state-of-the-art methods.
IEEE JOURNAL OF SELECTED TOPICS IN APPLIED EARTH OBSERVATIONS AND REMOTE SENSING
(2021)
Article
Engineering, Electrical & Electronic
Yongjie Zheng, Sicong Liu, Qian Du, Hui Zhao, Xiaohua Tong, Michele Dalponte
Summary: This article presents a novel multitemporal deep fusion network (MDFN) for short-term multitemporal HR images classification, which outperforms existing methods on remote sensing datasets and shows potential for more accurate land use/land cover mapping using short-term multitemporal HR images.
IEEE JOURNAL OF SELECTED TOPICS IN APPLIED EARTH OBSERVATIONS AND REMOTE SENSING
(2021)
Article
Engineering, Electrical & Electronic
Yingying Zhao, Guizhou Zheng, Zhangyan Xu, Zhonghang Qiu, Zhixing Chen
Summary: In this article, we propose a novel multiscale feature weighted-aggregating and boundary enhancement network (MFBE-Net) for the segmentation of high-resolution remote sensing images (HRRSIs). The proposed network effectively extracts objects of various sizes by weighted-integrating deep features, shallow features, and global information. The boundary enhancement module solves the problem of blurry boundary information and accurately locates its positions. Experimental results demonstrate that the proposed framework outperforms other mainstream deep learning methods and improves the accuracy of HRRSI segmentation.
IEEE JOURNAL OF SELECTED TOPICS IN APPLIED EARTH OBSERVATIONS AND REMOTE SENSING
(2022)
Article
Computer Science, Information Systems
Davit Marikyan, Savvas Papagiannidis, Eleftherios Alamanos
Summary: This study addresses the outcomes of technology use when it falls short of expectations and the coping mechanisms users may use in such circumstances. By adopting Cognitive Dissonance Theory, the study explores how negative disconfirmation of expectations can result in positive outcomes and how negative emotions impact the selection of dissonance reduction mechanisms. The study finds that post-disconfirmation dissonance leads to feelings of anger, guilt, and regret, which correlate with dissonance reduction mechanisms, ultimately affecting satisfaction and well-being.
INFORMATION SYSTEMS FRONTIERS
(2023)
Article
Computer Science, Artificial Intelligence
Kalifou Rene Traore, Andres Camero, Xiao Xiang Zhu
Summary: This study proposes a data-driven technique to initialize a population-based neural architecture search algorithm. The technique involves calibrated clustering analysis and extracting centroids for initialization. Experimental results show that compared to random and Latin hypercube sampling, this technique significantly improves performance in certain search scenarios and can be applied to improve search on other datasets.
ANNALS OF MATHEMATICS AND ARTIFICIAL INTELLIGENCE
(2023)
Article
Engineering, Electrical & Electronic
Rodrigo Capobianco Guido, Tulay Adali, Emil Bjoernson, Laure Blanc-Feraud, Ulisses Braga-Neto, Behnaz Ghoraani, Christian Jutten, Alle-Jan Van der Veen, Hong Vicky Zhao, Xiaoxing Zhu
Summary: The IEEE Signal Processing Society has provided 75 years of service to the signal processing community, making significant contributions to technological advancement.
IEEE SIGNAL PROCESSING MAGAZINE
(2023)
Article
Geography, Physical
Jianhua Guo, Qingsong Xu, Yue Zeng, Zhiheng Liu, Xiao Xiang Zhu
Summary: Urban tree canopy maps are crucial for providing urban ecosystem services. This study developed a semi-supervised deep learning method to robustly segment urban trees from high-resolution remote sensing images in order to better serve Brazil's urban ecosystem. The results showed that the urban tree canopy coverage in Brazil ranges from 5% to 35%, with an average coverage of approximately 18.68%. These canopy maps quantified the nationwide urban tree canopy inequality problem in Brazil. It is expected that these maps will encourage research on Brazilian urban ecosystem services, support urban development, and improve inhabitants' quality of life to achieve the Agenda for Sustainable Development goals.
ISPRS JOURNAL OF PHOTOGRAMMETRY AND REMOTE SENSING
(2023)
Article
Engineering, Electrical & Electronic
Lukas Kondmann, Sudipan Saha, Xiao Xiang Zhu
Summary: In this article, we explore the combination of unsupervised and supervised methods in a semisupervised framework to improve change detection performance. By using the unsupervised SiROC model to generate pseudolabels and selecting the most confident ones for pretraining different student models, we achieve robust results across various scenarios. The results show that pseudo-label pretraining produces significant performance gains, even when more labeled data is available. Moreover, the confidence selection of SiROC is effective and the performance gains are generalizable to unseen scenes.
IEEE JOURNAL OF SELECTED TOPICS IN APPLIED EARTH OBSERVATIONS AND REMOTE SENSING
(2023)
Article
Engineering, Electrical & Electronic
Milad Niroumand-Jadidi, Francesca Bovolo
Summary: A new deep-learning model called DOABLE-Net is proposed to detect and quantify harmful cyanobacterial blooms in inland waters. It leverages a large training set of R-rs data and provides more accurate and robust retrievals compared to the Pan-based method. The performance of DOABLE-Net is minimally impacted by the absence of a Pan band in Sentinel-2 data analysis.
IEEE JOURNAL OF SELECTED TOPICS IN APPLIED EARTH OBSERVATIONS AND REMOTE SENSING
(2023)
Article
Engineering, Electrical & Electronic
Parth Naik, Michele Dalponte, Lorenzo Bruzzone
Summary: This study proposes an automated machine learning framework for predicting forest above-ground biomass (AGB). The framework reduces human-bias through hyper-parameter optimization and automatic feature extraction, and improves prediction accuracy through model ensembling.
IEEE JOURNAL OF SELECTED TOPICS IN APPLIED EARTH OBSERVATIONS AND REMOTE SENSING
(2023)
Article
Geochemistry & Geophysics
Amos Bortiew, Swarnajyoti Patra, Lorenzo Bruzzone
Summary: This letter proposes a novel active learning technique for sparse representation classifiers (SRCs) that combines uncertainty and diversity criteria to design the query function. The proposed technique outperforms other state-of-the-art methods in terms of classification performance.
IEEE GEOSCIENCE AND REMOTE SENSING LETTERS
(2023)
Article
Engineering, Electrical & Electronic
Luca Bergamasco, Francesca Bovolo, Lorenzo Bruzzone
Summary: Multisensor data analysis utilizes heterogeneous data from multiple remote sensing systems to improve classification results. A supervised deep-learning method is proposed to analyze multiscale and multitemporal remote sensing images acquired by different sensors. The method processes high-resolution images with a residual network and analyzes spatial and temporal information using a 3-D convolutional neural network. The effectiveness of the method is demonstrated through experiments on two datasets.
IEEE JOURNAL OF SELECTED TOPICS IN APPLIED EARTH OBSERVATIONS AND REMOTE SENSING
(2023)
Article
Geochemistry & Geophysics
Qingsong Xu, Yilei Shi, Xin Yuan, Xiao Xiang Zhu
Summary: In this paper, a practical universal domain adaptation (UniDA) approach is proposed for remote sensing image scene classification, which requires no prior knowledge on the label sets. The proposed UniDA method distinguishes the shared and private label sets in each domain to promote adaptation and successfully recognize unknown samples. Empirical results demonstrate the effectiveness and practicality of the proposed model for remote sensing image scene classification, regardless of the availability of source data.
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING
(2023)
Article
Remote Sensing
Qingyu Li, Sebastian Krapf, Yilei Shi, Xiao Xiang Zhu
Summary: Promoting solar technology can provide affordable, reliable, and modern energy for all people while reducing energy-related emissions and pollutants, contributing to sustainable development goals. Aerial imagery offers a cost-effective approach for large-scale rooftop solar potential analysis compared to other data sources.
INTERNATIONAL JOURNAL OF APPLIED EARTH OBSERVATION AND GEOINFORMATION
(2023)
Article
Remote Sensing
Konrad Heidler, Lichao Mou, Di Hu, Pu Jin, Guangyao Li, Chuang Gan, Ji-Rong Wen, Xiao Xiang Zhu
Summary: Many deep learning approaches rely on pretrained backbone networks from large datasets, but the lack of such datasets in remote sensing is a challenge. We propose a label-free, self-supervised approach using imagery and audio correspondence to pretrain deep neural networks in remote sensing. We introduce the SoundingEarth dataset containing aerial imagery and crowd-sourced audio samples, and use it to pretrain ResNet models. Our approach outperforms existing pretraining strategies for remote sensing imagery. The dataset, code, and pretrained model weights are available at https://github.com/khdlr/SoundingEarth.
INTERNATIONAL JOURNAL OF APPLIED EARTH OBSERVATION AND GEOINFORMATION
(2023)
Article
Geochemistry & Geophysics
Yongjie Zheng, Sicong Liu, Lorenzo Bruzzone
Summary: This letter proposes a lightweight end-to-end attention-enhanced feature fusion network for hyperspectral image classification. The network effectively utilizes spectral-spatial information and achieves accurate classification even with few training samples. Experimental results demonstrate the superiority of the proposed approach compared to state-of-the-art deep learning methods.
IEEE GEOSCIENCE AND REMOTE SENSING LETTERS
(2023)
Article
Geochemistry & Geophysics
Lifeng Wang, Junguo Zhang, Liguo Wang, Lorenzo Bruzzone
Summary: This letter proposes an end-to-end semantic feature fused global learning framework for hyperspectral image multiclass change detection. The framework includes a global spatialwise fully convolutional network, a global hierarchical sampling strategy, a semantic-spatial feature fusion unit, and a semantic feature enhancement module.
IEEE GEOSCIENCE AND REMOTE SENSING LETTERS
(2023)
Article
Engineering, Electrical & Electronic
Daniel Carcereri, Paola Rizzoli, Dino Ienco, Lorenzo Bruzzone
Summary: This article presents a study on using deep learning to estimate forest height from InSAR data. The proposed fully convolutional neural network framework achieves good performance when tested on multiple sites, with an overall mean error of 1.46 m and mean absolute error of 4.2 m.
IEEE JOURNAL OF SELECTED TOPICS IN APPLIED EARTH OBSERVATIONS AND REMOTE SENSING
(2023)