YZYong Zhang

Yong Zhang

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Yong Zhang has a total of 40 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Top Cited Papers

    LMCNet: Lightweight Modality Compensation Network Via Knowledge Distillation for Salient Ship Detection Under Missing-Modality Conditions

    by Weibao Xue, Jiaqiu Ai, Yanan Zhu, Xinyu Sun, Yong Zhang, Gui Gao

    Salient ship detection is critical for maritime applications that require accurate localization. Although the fusion of Synthetic Aperture Radar (SAR) and Automatic Identification System (AIS) data has proven effective in enhancing saliency and suppressing background clutter, practical deployment faces two major limitations: the limited unavailability of AIS data and the high computational overhead of existing multimodal models. These limitations pose a fundamental challenge in balancing detection accuracy and efficiency under missing modality and resource-constrained environments. To address this issue, a Lightweight Modality Compensation Network (LMCNet) is proposed. A multimodal teacher network is trained with SAR and AIS inputs to learn rich and complementary representations. Meanwhile, a compact, single-modality student network that relies only on SAR is designed to support low-cost, real-time deployment. To enable robust knowledge compensation and transfer, this paper designs a knowledge distillation strategy consisting of three modules: structure-aware attention distillation for spatial alignment, cross-head teacher distillation for semantic enhancement, and adaptive loss scheduling for dynamic optimization. This unified design allows the student model to inherit spatial precision and semantic awareness from the teacher, achieving strong performance even with limited input modalities. Extensive experiments on two datasets show that our distilled student model improves Eξ E by 1.39% and Fwβ by 4.37% compared to state-of-the-art methods, demonstrating its superior balance of accuracy and efficiency in real-world deployment scenarios.


  • CSI-Based Location-Independent Human Activity Recognition by Contrast Between Dual Stream Fusion Features

    by Yujie Wang, Guangwei Yu, Yong Zhang, Dun Liu, Yang Zhang

    Due to the fact that Channel State Information (CSI) data contains activity and environmental information, the features of the same activity vary significantly across different locations. Existing CSI-based Human Activity Recognition (HAR) systems achieve high recognition accuracy at training locations through mechanisms such as transfer learning and few-shot learning when learning new activities. However, they struggle to maintain accurate activity recognition at other locations. In this paper, we propose a Contrastive Fusion feature-based Location-independent HAR System (CFLH) to address this issue. Unlike existing methods that simultaneously train feature extractor and fully-connected layer classifier, CFLH system decouples the training of the feature extractor and classifier. It only requires obtaining loss through contrastive learning at the feature level to optimize the feature extractor. CFLH system randomly scales activity signals in the temporal dimension to enrich intra and inter-class features across different locations, constructing positive samples. Using labels, samples from different activity categories are treated as negative samples to expand inter-class feature differences. For more effective activity feature extraction, CFLH system employs a two-tower transformer to extract temporal and channel-stream features. These two features are then fused into a dual-stream fusion feature using an Attention and Residual-based Fusion module (AR-Fusion). Experimental results show that using samples of 3 activities from 12 points to train the feature extractor, and adding samples of 3 new activities at training points to train the classifier, the highest recognition accuracy for 3 new and old activities at the testing location reaches 94.48% and 95.71%, respectively.


  • CSI-GLSTN: A Location-Independent CSI Human Activity Recognition Method Based on Spatio-Temporal and Channel Feature Fusion

    by Yong Zhang, Dingchao Wu, Yujie Wang, Yang Zhang, Guangkun Jiand, Jiaqiu Ai

    In the field of Human Activity Recognition (HAR) based on Wi-Fi signals, channel characteristics are susceptible to environmental influences, leading to significant feature variations among samples from different locations. Consequently, when existing model-based HAR systems attempt to recognize activities across different locations, the accuracy of pre-trained models often deteriorates. To address this challenge, a location-independent HAR system based Channel State Information(CSI) named Graph structure Learning based graph Saptio-temporal and Transformer Network(CSI-GLSTN) is proposed, which treats the sub-channels of Wi-Fi signals as nodes in a graph and constructs an initial graph structure through a graph structure learning layer. It employs a graph spatio-temporal network to fuse temporal and channel features of activity samples and map the fused features back to the initial graph structure. Through optimization and updates with activity samples from different locations, the final graph structure integrates both spatio-temporal and channel features of activities, characterizing different activity types. This achieves the decoupling of activity features from their respective locations, ultimately serving the goal of location-independent HAR. After training on samples from ten different locations, the system achieves an average recognition accuracy of 95.2% for six different types of human activities. Experimental results validate the system’s generalization performance in HAR across different locations.


  • An SAR Target Classification Algorithm Based on the Central Coordinate Attention Module

    by Jiaqiu Ai, Zheng Qu, Zhicheng Zhao, Yong Zhang, Jun Shi, Hao Yan

    Recent studies show that the attention mechanism can greatly improve the performance of SAR target classification. However, the existing attention models, such as Squeeze-and-Excitation (SE), Convolutional Block Attention Module (CBAM), Coordinate Attention (CA), emphasize the importance of the attention features of different channels, different columns or different rows, ignoring the importance of pixel-level feature in the attention feature maps. Moreover, these models ignore the center-distribution characteristic of the targets in SAR images, so the feature importance difference between the central target and the interference at the image borders cannot be well distinguished. As a consequence, they cannot achieve a high classification accuracy. In order to solve the above problems, this paper proposes a SAR target classification algorithm based on the Central Coordinate Attention Module (CCAM). CCAM highlights those important pixels in the feature maps through a pixel-level attention feature fusion strategy. Therefore, the importance difference of each pixel in the feature maps can be well distinguished. In addition, CCAM designs a center-importance weighting kernel, which highlights the center-distribution characteristic information of the SAR targets in the images, and weakens the interference at the image borders. Therefore, the feature importance difference between the central target and the interference at the image borders can be well distinguished. Finally, in order to achieve high efficiency, the proposed CCAM is incorporated into the convolutional neural network (CNN) for final SAR target classification. Undoubtedly, CCAM can greatly elevate the target feature representation completeness, thus improving the SAR target classification accuracy with a high efficiency. Experimental results on the MSTAR dataset verify the superiority and effectiveness of the proposed CCAM.


  • A Location-Independent Human Activity Recognition Method Based on CSI: System, Architecture, Implementation

    by Yong Zhang, Andong Cheng, Bin Chen, Yujie Wang, Lu Jia

    In the application of human activity recognition (HAR) based on channel state information (CSI), due to the high dynamic characteristics of wireless channel to different environments, the features of human activity samples in different locations are different. In addition, the existing CSI-based HAR approaches limit the extraction of activity features to the Euclidean space and ignores the rich relational information between samples, categories and locations, which result in insufficient generalization performance for location-independent HAR. To address this challenge, this paper proposes a CSI-based location-independent HAR system CSI-MTGN. The system represents the classification task under each training sample collection location (TSCL) as a task, which is composed of three interactive parts: sample hidden representation, activity features extraction based on hierarchical graph neural network (HGNN) and information exchange based on multi-task learning. The proposed system improves the sample hidden representation, which is benefit for activity feature extraction and classification. The HGNN is designed to express various relationship information between samples, categories and locations in the form of graph structure, and the classification task under each TSCL is constructed through data augmentation, so as to improve the knowledge understanding and inference capabilities of the recognition model. The multi-task learning is used to achieve implicit data augmentation by sharing parameters among tasks through soft parameter sharing, and improves the generalization performance of the system. To validate the performance of the proposed system, experiments were conducted in a hall and a conference room, where samples of 10 categories of activities under 7 TSCLs were used for training the system, and the HAR accuracy rates at any locations were 94.1% and 93.3%, respectively.


  • CSI-based cross-scene human activity recognition with incremental learning

    by Yong Zhang, Fei He, Yujie Wang, Dingchao Wu, Guangwei Yu