Article
Geochemistry & Geophysics
Xiaobo Liu, Xiang Liu, Haoran Dai, Xudong Kang, Antonio Plaza, Wenjie Zu
Summary: This article introduces a multiscale unsupervised architecture based on generative adversarial networks (GANs) for remote sensing image pansharpening called Mun-GAN. Mun-GAN achieves high-resolution fusion of remote sensing images through a multiscale feature extractor, a self-adaptation weighted fusion module, and a nest feature aggregation module. Experimental results compared with other methods demonstrate that Mun-GAN yields better fusion results.
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING
(2023)
Article
Mathematics
Yifei Zhang, Weipeng Li, Daling Wang, Shi Feng
Summary: In this paper, a novel Multi-scale Residual Generative Adversarial Network (MRGAN) based on unsupervised learning is proposed for image translation. It eliminates the dependence on paired training samples and introduces a multi-scale layered residual network to improve performance.
Article
Engineering, Electrical & Electronic
Jiahui Ni, Zhimin Shao, Zhongzhou Zhang, Mingzheng Hou, Jiliu Zhou, Leyuan Fang, Yi Zhang
Summary: Pansharpening aims to obtain a high-resolution multispectral image by fusing a low-resolution multispectral image with a panchromatic image. A novel unsupervised network called LDP-Net is proposed to effectively combine the spectral information of the LRMS image with the spatial information of the PAN image. Experimental results demonstrate that LDP-Net achieves promising performance in both qualitative visual effects and quantitative metrics.
IEEE JOURNAL OF SELECTED TOPICS IN APPLIED EARTH OBSERVATIONS AND REMOTE SENSING
(2022)
Article
Geochemistry & Geophysics
Jian Kang, Ruben Fernandez-Beltran, Puhong Duan, Sicong Liu, Antonio J. Plaza
Summary: The article presents a new unsupervised deep metric learning model called SauMoCo, designed to characterize unlabeled RS scenes by defining spatial augmentation criteria and constructing a queue of deep embeddings. The proposed approach substantially enhances the discrimination ability among complex land cover categories of RS tiles.
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING
(2021)
Article
Geochemistry & Geophysics
Anzhu Yu, Bing Liu, Xuefeng Cao, Chunping Qiu, Wenyue Guo, Yujun Quan
Summary: This research investigates a semi-supervised approach for building extraction from remotely sensed images. The backbone network is first trained using pixel-level self-supervised learning, and then combined with multiscale features for prediction. The experimental results show improvements in terms of intersection over union (IoU) and F1-score compared to supervised and instance-level SSL pretrained methods.
IEEE GEOSCIENCE AND REMOTE SENSING LETTERS
(2022)
Article
Environmental Sciences
Jun Chen, Weifeng Xu, Yang Yu, Chengli Peng, Wenping Gong
Summary: This paper proposes a reliable label-supervised pixel attention mechanism for building segmentation in UAV imagery. Experimental results demonstrate that the method outperforms previous weakly supervised methods on a UAV dataset.
Article
Engineering, Electrical & Electronic
Jieqiong Song, Jun Li, Hao Chen, Jiangjiang Wu
Summary: The article introduces an unsupervised domain mapping model based on adversarial learning for rapid map generation from remote sensing images. By using circularity-consistency and geometrical-consistency constraints, as well as an improved residual block Unet, the fidelity and geometry precision of the generated maps are enhanced. Experiments demonstrate that the model can efficiently generate maps on two distinct datasets and outperform other methods.
IEEE JOURNAL OF SELECTED TOPICS IN APPLIED EARTH OBSERVATIONS AND REMOTE SENSING
(2021)
Article
Computer Science, Information Systems
Menglei Zhang, Qiang Ling
Summary: The paper introduces a supervised pixel-wise Generative Adversarial Network (SPGAN) that can upscale low-resolution face images to larger versions with multiple scaling factors. By utilizing face features and identity prior, SPGAN enhances face recognition performance by focusing on texture details. Extensive experiments show that SPGAN produces more photo-realistic super-resolution images and better face recognition accuracy compared to state-of-the-art methods.
IEEE TRANSACTIONS ON MULTIMEDIA
(2021)
Article
Computer Science, Interdisciplinary Applications
Hanxiang Wang, Yanfen Li, L. Minh Dang, Sujin Lee, Hyeonjoon Moon
Summary: An innovative framework combining weakly supervised learning methods and fully supervised learning methods is proposed for crack detection and segmentation in tunnel images. The proposed method successfully judges the risk levels of detected cracks and performs calculations on different types of cracks. Furthermore, the framework achieves comparable performance to manual annotation-based frameworks, showcasing the effectiveness of weakly supervised learning methods in accurately labeling images.
COMPUTERS IN INDUSTRY
(2021)
Article
Computer Science, Artificial Intelligence
Marcus de Carvalho, Mahardhika Pratama, Jie Zhang, Edward Yapp Kien Yee
Summary: This article discusses the problem of online unsupervised cross-domain adaptation and proposes the ACDC framework. It effectively addresses the challenges in data streams through a self-evolving neural network structure and achieves promising results in experiments.
KNOWLEDGE-BASED SYSTEMS
(2022)
Article
Geochemistry & Geophysics
Siyuan Wang, Dongyang Hou, Huaqiao Xing
Summary: In this article, a new self-supervised-driven open-set UDA (SSOUDA) method is proposed for optical remote sensing scene classification and retrieval, which combines contrastive self-supervised learning with consistency self-training (CST). The method utilizes a contrastive self-supervised learning network to learn discriminative features from unlabeled target domain data, and a novel open-set class learning module is developed based on two-level confidence rules and the consistency self-training strategy. Experimental results demonstrate that the proposed method achieves superior performances on six open-set cross-domain scenarios.
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING
(2023)
Article
Geochemistry & Geophysics
Asad Mahmood, Michael Sears
Summary: Modeling the underlying noise in hyperspectral images can provide important insights into sensor characteristics and image processing. The proposed method for per-pixel noise estimation, capable of handling spectral correlation in noise, outperforms existing methods in cases with noise correlation, as demonstrated by simulation results.
IEEE GEOSCIENCE AND REMOTE SENSING LETTERS
(2022)
Article
Computer Science, Artificial Intelligence
Yu Wang, Jingyang Lin, Qi Cai, Yingwei Pan, Ting Yao, Hongyang Chao, Tao Mei
Summary: Unsupervised learning is at the tipping point and contrastive learning has led to state-of-the-art performance. This paper introduces a novel probabilistic graphical model called LORAC that incorporates the low rank promoting prior into contrastive learning. The algorithm surpasses existing methods on multiple benchmarks, including image classification, object detection, instance segmentation, and keypoint detection.
IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE
(2023)
Article
Automation & Control Systems
Yongchao Zhang, Zhaohui Ren, Shihua Zhou, Ke Feng, Kun Yu, Zheng Liu
Summary: The fault diagnosis of rolling bearing is crucial for production efficiency and safety. In this paper, a supervised contrastive learning-based domain adaptation network (SCLDAN) is proposed to improve the accuracy of fault diagnosis by learning raw signal features and achieving global domain alignment.
IEEE-ASME TRANSACTIONS ON MECHATRONICS
(2022)
Review
Biochemical Research Methods
Mohamed Nadif, Francois Role
Summary: Biomedical scientific literature is growing rapidly, making it challenging to identify relevant results; automated information extraction tools based on text mining techniques are essential; deep neural networks have significantly advanced this research field.
BRIEFINGS IN BIOINFORMATICS
(2021)