Squeeze-SwinFormer: Spectral Squeeze and Excitation Swin Transformer Network for Hyperspectral Image Classification

Ullah F. Ullah I. Khan K. Khan S. Wang Q. Algamdi S.A. Aldossary H.
2025 Institute of Electrical and Electronics Engineers Inc.

IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
2025 #18 21400 - 21418 pp.

Hyperspectral images are highly complex, containing a richer spectral dimension compared to conventional images. Deep learning methods are increasingly being applied to process this three-dimensional data for hyperspectral image classification (HSIC). The vision transformer model is steadily gaining prominence in computer vision, emerging as a potential alternative to traditional convolutional neural network (CNN) architectures. Although transformers are powerful due to their self-attention mechanisms, they face challenges related to scalability and efficiency, particularly when processing high-resolution images. In order to solve these problems, in this article, we propose a novel Squeeze-SwinFormer network for HSIC. This novel approach integrates spectral squeeze and excitation (SE) blocks into the shifted window (Swin) transformer architecture to enhance feature extraction and attention mechanisms in the model. The SE block dynamically recalibrates channel-wise feature responses, improving the models focus on significant features. The Squeeze-SwinFormer, equipped with window-based multihead self-attention, efficiently captures long-range dependencies with reduced computational complexity by partitioning input data into nonoverlapping windows. In addition, the SE blocks integration within the Swin transformer enables improved global contextual understanding by selectively weighting the feature maps. We further refine the architecture using various components, such as patch extraction and embedding layers, and a patch merging strategy, ensuring efficient multiscale feature extraction. The proposed architecture demonstrates superior performance in tasks requiring attention to intricate details in HSIC. The comprehensive experimental results demonstrate that our proposed model achieved higher test accuracies against the state-of-the-art method, with results of 99.93%, 98.31%, 99.50%, and 97.75% on the Salinas, Indian Pines, University of Pavia, and Kennedy Space Center public benchmark HSIC datasets, respectively. Our approach also shows good generalization ability when applied to new datasets. Overall, our proposed approach represents a promising direction for HSIC.

Classification , hyperspectral image classification (HSIC) , squeeze and excitation (SE) , Squeeze-SwinFormer , Swin transformer (SwinT)

Text of the article Перейти на текст статьи

Wuxi University, School of Internet of Things Engineering, Jiangsu Foreign Expert Lab, Wuxi University School of Internet of Things Engineering Foreign Expert Lab, Wuxi, 214105, China
Chengdu University of Technology, School of Computer Science, Sichuan, 611730, China
Nazarbayev University, Department of Computer Science, School of Engineering and Digital Sciences, Astana, 010000, Kazakhstan
Qilu Institute of Technology, School of Computer and Information Engineering, Jinan, 250202, China
Prince Sattam Bin Abdulaziz University, Department of Software Engineering, College of Computer Science and Engineering, Al Kharj, 16278, Saudi Arabia
Imam Abdulrahman Bin Faisal University, Computer Science Department, College of Science and Humanities, Jubail, 31961, Saudi Arabia

Wuxi University
Chengdu University of Technology
Nazarbayev University
Qilu Institute of Technology
Prince Sattam Bin Abdulaziz University
Imam Abdulrahman Bin Faisal University

10 лет помогаем публиковать статьи Международный издатель

Книга Публикация научной статьи Волощук 2026 Book Publication of a scientific article 2026