Wangzhi Xing

Wangzhi Xing

Auto-generated profile, based on publicly available information

Wangzhi Xing

Department of Computer Science and Mathematics, Bangladesh Agricultural University, Mymensingh, Bangladesh

Research Focus & Metrics

Academic Career Length

3Years (2024-Present)

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

Representative Publications

View All Publications
24 June 2024 in IEEE Transactions on Geoscience and Remote Sensing
Verified
Journal Article
Closed access
MIMohammad Aminul Islam
Mohammad Aminul Islam
WXWangzhi Xing
Wangzhi Xing
JZJun Zhou
Jun Zhou
YGYongsheng Gao
Yongsheng Gao
KPKuldip K. Paliwal
Kuldip K. Paliwal
Hyperspectral images, with their many spectral bands, provide a rich source of material information about an object that can be effectively used for object tracking. However, many trackers in this domain rely on detection-based techniques, which often perform suboptimally in challenging scenarios such as managing occlusions and distinguishing objects in cluttered backgrounds. This underperformance is primarily due to the presence of multiple spectral bands and the inability to leverage this abundance of data for effective tracking. Additionally, the scarcity of annotated hyperspectral videos and the absence of comprehensive temporal information exacerbate these difficulties, further limiting the effectiveness of current tracking methods. To address these challenges, this paper introduces the novel Hy-Tracker framework, designed to bridge the gap between hyperspectral data and state-of-the-art object detection methods. Our approach leverages the strengths of YOLOv7 for object tracking in hyperspectral videos, enhancing both accuracy and robustness in complex scenarios. The Hy-Tracker framework comprises two key components. We introduce a hierarchical attention band selection (HAS-BS) that selectively processes and groups the most informative spectral bands, thereby significantly improving detection accuracy. Additionally, we have developed a refined tracker that refines the initial detections by incorporating a classifier and a temporal network using gated recurrent units (GRU). The classifier distinguishes similar objects, while the temporal network models temporal dependencies across frames for robust performance despite occlusions and scale variations. Experimental results on hyperspectral benchmark datasets demonstrate the effectiveness of Hy-Tracker in accurately tracking objects across frames and overcoming the challenges inherent in detection-based hyperspectral object tracking.

19 May 2025 in IEEE Transactions on Geoscience and Remote Sensing
Verified
Journal Article
Closed access
MIMohammad Aminul Islam
Mohammad Aminul Islam
JZJun Zhou
Jun Zhou
WXWangzhi Xing
Wangzhi Xing
YGYongsheng Gao
Yongsheng Gao
KPKuldip K. Paliwal
Kuldip K. Paliwal
Hyperspectral object tracking is notably challenging due to the high-dimensional nature of the data and the necessity of seamlessly integrating spectral, spatial and temporal information. Traditional methods often emphasize detection-based or tracking-based networks, each leveraging their inherent strengths but overlooking the potential advantages of a combined approach, leading to suboptimal performance in complex, real-world scenarios. Furthermore, this challenge is amplified by the variability of spectral bands across datasets, making the maintenance of consistent tracking performance complicated. To address these issues, we propose a novel, unified approach that merges adaptive band selection with a multi-model ensemble strategy. We introduce a local and global attention-based unified band selection (UBS) technique that identifies the most informative three bands from any dataset, significantly reducing data complexity while preserving critical spectral and spatial information. This UBS method employs spectral independence, allowing it to process hyperspectral video frames with any number of bands as input, ultimately generating a three-band pseudocolor image. This is coupled with a multi-model ensemble framework, utilizing a local and global attention-based appearance module. The module selects the optimal candidate by computing the similarity between the proposals generated by the base models and historical frames. Experimental results show that our approach, UBSTrack, achieves state-of-the-art performance, delivering robust and accurate tracking under different real-world challenging scenarios. The code of UBSTrack is available at the following link: source code.

4 July 2026 in Artificial Intelligence Review
Verified
Journal Article
Open access
MIMohammad Aminul Islam
Mohammad Aminul Islam
FXFengchao Xiong
Fengchao Xiong
JZJun Zhou
Jun Zhou
WXWangzhi Xing
Wangzhi Xing
DBDong Bao
Dong Bao
YGYongsheng Gao
Yongsheng Gao
Visual object tracking is a fundamental problem in computer vision. Traditional tracking methods, which primarily rely on RGB imagery, often face difficulties in complex scenarios such as low resolution and background clutter. Hyperspectral imaging, which captures both spatial and spectral information across multiple narrow spectral bands, has emerged as a promising solution. However, hyperspectral tracking suffers from challenges including the complexity of spatial-spectral-temporal modeling, the high dimensionality of spectral bands, low spatial resolution, and limited training samples. In response, researchers have developed a variety of hyperspectral object tracking (HOT) methods, integrating innovations in band selection and advanced architectures such as Siamese networks, Transformers, and multimodal fusion-based approaches. Despite this progress, existing reviews primarily focus on specific sub-domains and provide limited coverage of recent developments in hyperspectral video camera technology and the associated data processing pipelines, including calibration and correction techniques. Therefore, this review aims to fill that gap by providing an in-depth and systematic overview of HOT. It examines the motivations driving this research, highlights recent technological and methodological developments, surveys publicly available datasets, and evaluates performance metrics. Finally, we discuss future research directions and prospects, offering a valuable resource for researchers and practitioners. Moreover, to support ongoing innovation, we have established a dedicated GitHub repository to document and continuously update relevant works: https://github.com/aamin0102/HOT-Review .

11 September 2026 in SSRN Electronic Journal
Verified
Preprint
Open access
WXWangzhi Xing
Wangzhi Xing
DCDiqi Chen
Diqi Chen
MIMohammad Aminul Islam
Mohammad Aminul Islam
JZJun Zhou
Jun Zhou
Infrared-visible image fusion (IVIF) becomes challenging when trustworthy visible information and locally degraded responses coexist within a single image. Texture and scene appearance may remain useful in some regions, whereas other regions exhibit low-light darkness, low contrast, glare-like washout, or saturation/overexposure and therefore require selective treatment during fusion. Existing approaches to degraded IVIF inputs improve robustness, but do not fully explore how local trustworthiness should coordinaterepresentation learning, information transfer, and supervision. To address this gap, we propose RePrior, a framework based on reliability-anchored factorized priors. Weak regional cues anchor reliable and unreliable components with a two-component Gaussian mixture prior over source-specific visible features. Simultaneously, joint factorized-prior matching regularizes the organization of modality-common and source-specific representations. The resulting soft unreliable-component assignment drives two distinct controls: a strict gate for visible-detail transfer and a more tolerant gate for appearance preservation. This design suppresses degradation-dominated responses without uniformly discarding valid smooth regions. Across four IVIF benchmarks, RePrior ranks among the top two in 15 of 16 dataset--metric comparisons against 11 representative methods. Mechanism-oriented ablations support the intended prior semantics,shared/private latent organization, and role-specific control, whiledownstream perception evaluations further support practical utility. Inseparately trained secondary studies, the same reliability-conditionedscaffold is adapted to multi-exposure fusion (MEF), multi-focus fusion (MFF),and medical image fusion (MIF) by defining reliability as local exposureusefulness, focus/sharpness confidence, and local information confidence,respectively, yielding top-two results on four of five metrics in each task.