Return
Multimodal semantic-scale network for remote sensing image classification
DOI:10.1016/j.inffus.2026.104361.png)
Abstract
En 中文
• Combines convolutions with a Euclidean-based semantic self-attention mechanism. • Scale-adaptive module that balances dimensionality reduction and feature recovery. • Lightweight high-performance architecture is suitable for practical applications.
Keywords:
convolution
semantic self-attention
scale-adaptive module
lightweight architecture
remote sensing image classification

