MS-SSE-Net: A Multi-Scale Spatial Squeeze-and-Excitation Network for Structural Damage Detection in Civil and Geotechnical Engineering

2026-04-16 · Source: Computer Vision and Pattern Recognition · Field: Construction & Real Estate — Construction Technology & Building, Infrastructure & Civil Engineering, Artificial Intelligence & Machine Learning · Depth: Expert, quick

Summary

MS-SSE-Net is a new deep learning framework designed for structural damage classification in civil and geotechnical engineering, addressing challenges posed by varied damage patterns and environmental conditions. Built on a DenseNet201 backbone, the model incorporates multi-scale feature extraction using parallel depthwise convolutions to capture both local and contextual features. It integrates squeeze-and-excitation style channel attention and spatial attention mechanisms to highlight informative regions and mitigate noise. After feature refinement, global average pooling and a fully connected layer generate predictions. Evaluated on the StructDamage dataset, MS-SSE-Net achieved 99.31% precision, 99.25% recall, 99.27% F1-score, and 99.26% accuracy, outperforming the baseline DenseNet201's 98.62% precision, 98.53% recall, 98.58% F1-score, and 98.53% accuracy.

Key takeaway

For civil and geotechnical engineers developing automated structural inspection systems, MS-SSE-Net offers a robust deep learning approach to improve damage classification accuracy. Its integration of multi-scale feature extraction and attention mechanisms can significantly enhance the reliability of damage detection, potentially reducing false positives and improving maintenance scheduling. Consider adopting similar architectural components to boost the performance of your existing image-based damage assessment models.

Key insights

MS-SSE-Net enhances structural damage detection via multi-scale feature extraction and attention mechanisms.

Principles

• Multi-scale features improve damage pattern recognition.
• Attention mechanisms enhance feature relevance.

Method

MS-SSE-Net uses parallel depthwise convolutions for multi-scale feature extraction, followed by channel and spatial attention, global average pooling, and a fully connected layer for classification.

In practice

• Apply depthwise convolutions for multi-scale feature capture.
• Integrate Squeeze-and-Excitation for channel attention.

Topics

• MS-SSE-Net
• Structural Damage Detection
• Deep Learning
• Multi-Scale Feature Extraction
• Spatial Squeeze-and-Excitation

Best for: AI Scientist, Computer Vision Engineer, Research Scientist

Related on AIssential

• DeltaSeg: Tiered Attention and Deep Delta Learning for Multi-Class Structural Defect Segmentation — Tiered attention and refined skip connections significantly improve structural defect segmentation accuracy and generalization.
• Bridging Spatial And Frequency Views For Disaster Assessment: Benefits And Limitations — Combining spatial and frequency domain features measurably improves deep learning models for disaster damage assessment.
• Balancing Real and Synthetic Data for CNN-based Masonry Crack Detection — Synthetic data can significantly reduce real data collection needs for CNN-based crack detection while maintaining or improving accuracy.
• Multi-Modal Attention for Automated Disaster Damage Assessment Using Remote Sensing Imagery and Deep Learning — Multi-modal attention effectively fuses bi-temporal remote sensing data for automated disaster damage classification.
• Three-dimensional Damage Visualization of Civil Structures via Gaussian Splatting-enabled Digital Twins — Gaussian Splatting enhances 3D damage visualization in civil infrastructure digital twins, improving accuracy and efficiency.
• SSH-Net: A Deep Neural Network for Predicting Failure Time Distribution Functions under Competing Risks with Application to GPU Data — SSH-Net improves competing risks survival analysis by structuring neural networks to data hierarchies, enhancing prediction and interpretability.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Computer Vision and Pattern Recognition.