LSNet: Extremely Light-Weight Siamese Network For Change Detection in Remote Sensing Image

TL;DR

This paper introduces LSNet, a highly efficient Siamese network for remote sensing image change detection that significantly reduces model size and computation while maintaining high accuracy.

Contribution

It proposes a lightweight architecture using depthwise separable atrous convolution and streamlined feature fusion to improve efficiency in RSI change detection.

Findings

Parameters reduced by 90.35%

Computation reduced by 91.34%

Accuracy drops only 1.5%

Abstract

The Siamese network is becoming the mainstream in change detection of remote sensing images (RSI). However, in recent years, the development of more complicated structure, module and training processe has resulted in the cumbersome model, which hampers their application in large-scale RSI processing. To this end, this paper proposes an extremely lightweight Siamese network (LSNet) for RSI change detection, which replaces standard convolution with depthwise separable atrous convolution, and removes redundant dense connections, retaining only valid feature flows while performing Siamese feature fusion, greatly compressing parameters and computation amount. Compared with the first-place model on the CCD dataset, the parameters and the computation amount of LSNet is greatly reduced by 90.35\% and 91.34\% respectively, with only a 1.5\% drops in accuracy.