Xiaowei Zhou

Dr. Xiaowei Zhou


Research Focus & Metrics

Academic Career Length

6Years (2021-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
28 January 2025 in IEEE Transactions on Geoscience and Remote Sensing
Verified
Journal Article
Closed access
Feng GaoFeng Gao
Feng Gao
XJXuepeng Jin
Xuepeng Jin
XZXiaowei Zhou
Xiaowei Zhou
Junyu DongJunyu Dong
Junyu Dong
Qian DuQian Du
Qian Du
In the field of multi-source remote sensing image classification, remarkable progress has been made by using Convolutional Neural Network (CNN) and Transformer. While CNNs are constrained by their local receptive fields, Transformers mitigate this issue with their global attention mechanism. However, Transformers come with the trade-off of higher computational complexity. Recently, Mamba-based methods built upon the State Space Model (SSM) have shown great potential for long-range dependency modeling with linear complexity, but they have rarely been explored for multi-source remote sensing image classification tasks. To address this issue, we propose the Multi-Scale Feature Fusion Mamba (MSFMamba) network, a novel framework designed for the joint classification of hyperspectral image (HSI) and Light Detection and Ranging (LiDAR)/Synthetic Aperture Radar (SAR) data. The MSFMamba network is composed of three key components: the Multi-Scale Spatial Mamba (MSpa-Mamba) block, the Spectral Mamba (Spe-Mamba) block, and the Fusion Mamba (Fus-Mamba) block. The MSpa-Mamba block employs a multi-scale strategy to reduce computational cost and alleviate feature redundancy in multiple scanning routes, ensuring efficient spatial feature modeling. The Spe-Mamba block focuses on spectral feature extraction, addressing the unique challenges of HSI data representation. Finally, the Fus-Mamba block bridges the heterogeneous gap between HSI and LiDAR/SAR data by extending the original Mamba architecture to accommodate dual inputs, enhancing cross-modal feature interactions and enabling seamless data fusion. Together, these components enable MSFMamba to effectively tackle the challenges of multi-source data classification, delivering improved performance with optimized computational efficiency. Comprehensive experiments on four real-world multi-source remote sensing datasets (Berlin, Augsburg, Houston2018, and Houston2013) demonstrate the superiority of MSFMamba outperforms several state-of-the-art methods and achieves overall accuracies of 76.92%, 91.38%, 92.38%, and 92.86%, respectively. The source codes of MSFMamba will be publicly available at https://github.com/oucailab/MSFMamba.

22 July 2024 in IEEE Geoscience and Remote Sensing Letters
Verified
Journal Article
Closed access
JXJiangwei Xie
Jiangwei Xie
Feng GaoFeng Gao
Feng Gao
XZXiaowei Zhou
Xiaowei Zhou
Junyu DongJunyu Dong
Junyu Dong
Synthetic aperture radar (SAR) image change detection is critical in remote sensing image analysis. Recently, the attention mechanism has been widely used in change detection tasks. However, existing attention mechanisms often employ down-sampling operations such as average pooling on the Key and Value components to enhance computational efficiency. These irreversible operations result in the loss of high-frequency components and other important information. To address this limitation, we develop Wavelet-based Bi-dimensional Aggregation Network (WBANet) for SAR image change detection. We design a wavelet-based self-attention block that includes discrete wavelet transform and inverse discrete wavelet transform operations on Key and Value components. Hence, the feature undergoes downsampling without any loss of information, while simultaneously enhancing local contextual awareness through an expanded receptive field. Additionally, we have incorporated a bi-dimensional aggregation module that boosts the non-linear representation capability by merging spatial and channel information via broadcast mechanism. Experimental results on three SAR datasets demonstrate that our WBANet significantly outperforms contemporary state-of-the-art methods. Specifically, our WBANet achieves 98.33%, 96.65%, and 96.62% of percentage of correct classification (PCC) on the respective datasets, highlighting its superior performance. Source codes will be available at https: //github.com/summitgao/WBANet.

12 May 2025 in IEEE Transactions on Geoscience and Remote Sensing
Verified
Journal Article
Closed access
Feng GaoFeng Gao
Feng Gao
SLSheng Liu
Sheng Liu
CGChuanzheng Gong
Chuanzheng Gong
XZXiaowei Zhou
Xiaowei Zhou
JWJiayi Wang
Jiayi Wang
Qian DuQian Du
Qian Du
Multisource remote sensing data joint classification aims to provide accuracy and reliability of land-cover classification by leveraging the complementary information from multiple data sources. Existing methods confront two challenges: interfrequency multisource feature coupling and inconsistency of complementary information exploration. To solve these issues, we present a prototype-based information compensation network (PICNet) for land-cover classification based on hyperspectral image (HSI) and synthetic aperture radar (SAR)/light detection and ranging (LiDAR) data. Specifically, we first design a frequency interaction module (FIM) to enhance the interfrequency coupling in multisource feature extraction. The multisource features are first decoupled into high- and low-frequency components. Then, these features are recoupled to achieve efficient interfrequency communication. Afterward, we design a prototype-based information compensation module (PICM) to model the global multisource complementary information. Two sets of learnable modality prototypes are introduced to represent the global modality information of multisource data. Subsequently, cross-modal feature integration and alignment are achieved through cross-attention computation between the modality-specific prototype vectors and the raw feature representations. Extensive experiments on three public datasets demonstrate the significant superiority of our PICNet over state-of-the-art methods. The codes are available at https://github.com/oucailab/PICNet

Verified
Journal Article
Open access
PDF file
YHYang Hong
Yang Hong
XZXiaowei Zhou
Xiaowei Zhou
RHRuzhuang Hua
Ruzhuang Hua
QLqingxuan Lv
qingxuan Lv
Junyu DongJunyu Dong
Junyu Dong
Object segmentation, a key type of image segmentation, focuses on detecting and delineating individual objects within an image, essential for applications like robotic vision and augmented reality. Despite advancements in deep learning improving object segmentation, underwater object segmentation remains challenging due to unique underwater complexities such as turbulence diffusion, light absorption, noise, low contrast, uneven illumination, and intricate backgrounds. The scarcity of underwater datasets further complicates these challenges. The Segment Anything Model (SAM) has shown potential in addressing these issues, but its adaptation for underwater environments, AquaSAM, requires fine-tuning all parameters, demanding more labeled data and high computational costs. In this paper, we propose WaterSAM, an adapted model for underwater object segmentation. Inspired by Low-Rank Adaptation (LoRA), WaterSAM incorporates trainable rank decomposition matrices into the Transformer’s layers, specifically enhancing the image encoder. This approach significantly reduces the number of trainable parameters to 6.7% of SAM’s parameters, lowering computational costs. We validated WaterSAM on three underwater image datasets: COD10K, SUIM, and UIIS. Results demonstrate that WaterSAM significantly outperforms pre-trained SAM in underwater segmentation tasks, contributing to advancements in marine biology, underwater archaeology, and environmental monitoring.

24 April 2025 in Applied Sciences
Verified
Journal Article
Open access
PDF file
文于文雅 于
文雅 于
XZXiaowei Zhou
Xiaowei Zhou
DWDongSheng Wang
DongSheng Wang
Junyu DongJunyu Dong
Junyu Dong
With rapid urbanization exacerbating the challenges in resource allocation, environmental sustainability, and infrastructure management, City Information Modeling (CIM) has emerged as an indispensable digital solution for smart city development. CIM represents an advanced urban management paradigm that integrates Geographic Information Systems (GISs), Building Information Modeling (BIM), and the Internet of Things (IoT) to establish a multidimensional digital framework for comprehensive urban data management and intelligent decision making. While the existing research has primarily focused on technical architectures, governance models, and application scenarios, a systematic exploration of CIM’s data-driven characteristics remains limited. This paper reviews the evolution of CIM from a data-centric view introducing a research framework that systematically examines the data lifecycle, including acquisition, processing, analysis, and decision support. Furthermore, it explores the application of CIM in key areas such as smart transportation and digital twin cities, emphasizing its deep integration with big data, artificial intelligence (AI), and cloud computing to enhance urban governance and intelligent services. Despite its advancements, CIM faces critical challenges, including data security, privacy protection, and cross-sectoral data sharing. This survey highlights these limitations and points out the future research directions, including adaptive data infrastructure, ethical frameworks for urban data governance, intelligent decision-making systems leveraging multi-source heterogeneous data, and the integration of CIM with emerging technologies such as AI and blockchain. These innovations will enhance CIM’s capacity to support intelligent, resilient, and sustainable urban development. By establishing a theoretical foundation for CIM as a data-intensive framework, this survey provides valuable insights and forward-looking guidance for its continued research and practical implementation.

9 September 2022 in Machine Learning
Verified
Journal Article
Open access
PDF file
XZXiaowei Zhou
Xiaowei Zhou
ITIvor Tsang
Ivor Tsang
JYJie Yin
Jie Yin
Deep Neural Networks (DNNs) have recently achieved great success in many classification tasks. Unfortunately, they are vulnerable to adversarial attacks that generate adversarial examples with a small perturbation to fool DNN models, especially in model sharing scenarios. Adversarial training is proved to be the most effective strategy that injects adversarial examples into model training to improve the robustness of DNN models against adversarial attacks. However, adversarial training based on the existing adversarial examples fails to generalize well to standard, unperturbed test data. To achieve a better trade-off between standard accuracy and adversarial robustness, we propose a novel adversarial training framework called LAtent bounDary-guided aDvErsarial tRaining (LADDER) that adversarially trains DNN models on latent boundary-guided adversarial examples. As opposed to most of the existing methods that generate adversarial examples in the input space, LADDER generates a myriad of high-quality adversarial examples through adding perturbations to latent features. The perturbations are made along the normal of the decision boundary constructed by an SVM with an attention mechanism. We analyze the merits of our generated boundary-guided adversarial examples from a boundary field perspective and visualization view. Extensive experiments and detailed analysis on MNIST, SVHN, CelebA, and CIFAR-10 validate the effectiveness of LADDER in achieving a better trade-off between standard accuracy and adversarial robustness as compared with vanilla DNNs and competitive baselines.