Abstract: In hyperspectral image (HSI) classification, Transformer and CNN are widely used because they complement each other in extracting features. Nevertheless, existing Transformer-based methods ...

Abstract: To address the shortcomings of traditional deep learning applications for multi-label satellite image classification, this work shifts focus toward lightweight Vision Transformers (ViTs) ...

Gemini 3 Flash adds active vision with Python code execution, lifting accuracy by 5 to 10%, so you can trust verified results.