A Robust and Low Complexity Deep Learning Model for Remote Sensing Image Classification

TL;DR

This paper introduces a robust, low-complexity deep learning model for remote sensing image classification, utilizing efficient neural networks, attention mechanisms, and quantization to achieve competitive performance suitable for edge devices.

Contribution

It evaluates and enhances lightweight neural networks with attention and quantization, creating a practical model for remote sensing classification.

Findings

Achieved high accuracy on NWPU-RESISC45 dataset

Model size limited to 20 MB with quantization

Performance comparable to state-of-the-art methods

Abstract

In this paper, we present a robust and low complexity deep learning model for Remote Sensing Image Classification (RSIC), the task of identifying the scene of a remote sensing image. In particular, we firstly evaluate different low complexity and benchmark deep neural networks: MobileNetV1, MobileNetV2, NASNetMobile, and EfficientNetB0, which present the number of trainable parameters lower than 5 Million (M). After indicating best network architecture, we further improve the network performance by applying attention schemes to multiple feature maps extracted from middle layers of the network. To deal with the issue of increasing the model footprint as using attention schemes, we apply the quantization technique to satisfy the maximum of 20 MB memory occupation. By conducting extensive experiments on the benchmark datasets NWPU-RESISC45, we achieve a robust and low-complexity model, which is very competitive to the state-of-the-art systems and potential for real-life applications on edge devices.