MSCrackMamba: Leveraging Vision Mamba for Crack Detection in Fused Multispectral Imagery

TL;DR

This paper introduces MSCrackMamba, a novel two-stage approach combining super-resolution and Vision Mamba architecture to improve crack detection in fused multispectral images, addressing resolution mismatch and enhancing detection accuracy.

Contribution

It proposes a new two-stage paradigm using super-resolution and Vision Mamba for better multispectral crack detection, outperforming existing methods on a large dataset.

Findings

Achieved 3.55% higher mIoU than baseline methods.

Effectively aligns IR and RGB channels through super-resolution.

Demonstrates improved crack detection accuracy on Crack900 dataset.

Abstract

Crack detection is a critical task in structural health monitoring, aimed at assessing the structural integrity of bridges, buildings, and roads to prevent potential failures. Vision-based crack detection has become the mainstream approach due to its ease of implementation and effectiveness. Fusing infrared (IR) channels with red, green and blue (RGB) channels can enhance feature representation and thus improve crack detection. However, IR and RGB channels often differ in resolution. To align them, higher-resolution RGB images typically need to be downsampled to match the IR image resolution, which leads to the loss of fine details. Moreover, crack detection performance is restricted by the limited receptive fields and high computational complexity of traditional image segmentation networks. Inspired by the recently proposed Mamba neural architecture, this study introduces a two-stage paradigm called MSCrackMamba, which leverages Vision Mamba along with a super-resolution network to address these challenges. Specifically, to align IR and RGB channels, we first apply super-resolution to IR channels to match the resolution of RGB channels for data fusion. Vision Mamba is then adopted as the backbone network, while UperNet is employed as the decoder for crack detection. Our approach is validated on the large-scale Crack Detection dataset Crack900, demonstrating an improvement of 3.55% in mIoU compared to the best-performing baseline methods.