MixerSENet: A Lightweight Framework for Efficient Hyperspectral Image Classification

doi:10.1109/LGRS.2025.3616338

MixerSENet: A Lightweight Framework for Efficient Hyperspectral Image Classification

2025-01-01

0

PRE

AI

DOI：10.1109/LGRS.2025.3616338

Original

Original request for help

Save

Abstract

En 中文

In this letter, a novel framework, MixerSENet, is introduced for hyperspectral image (HSI) classification, designed to address the challenges of computational efficiency and limited labeled data. The proposed model processes HSI patches while maintaining consistent size and resolution throughout the network, effectively decoupling the mixing of spatial and channel dimensions. Notably, the MixerSENet is lightweight and computationally efficient, requiring fewer parameters compared to traditional models, making it suitable for resource-constrained environments. A squeeze-and-excitation (SE) block is incorporated into the model to refine feature extraction, enhancing the network’s ability to capture more informative features. Experimental results on two benchmark datasets demonstrate that the MixerSENet achieves superior performance, reaching an overall accuracy (OA) of 82.47% on the Houston13 dataset and 96.70% on the Qingyun dataset, outperforming state-of-the-art methods including the 3D-CNN, Kolmogorov–Arnold network (HybridKAN), HSIFormer, SimPoolFormer, and MorphMamba. Furthermore, a detailed analysis of computational efficiency shows that the MixerSENet achieves a favorable balance between accuracy and efficiency, with only 53 146 parameters and a low inference time, confirming its practicality for real-world applications. At publication, the source code will be publicly available at https://github.com/mqalkhatib/MixerSENet

Keywords:

Attention block

depth-wise convolution

hyperspectral imaging (HSI) classification

mixer networks

MixerSENet: A Lightweight Framework for Efficient Hyperspectral Image Classification

Abstract

Journal

Organization