FORTRESS: Function-composition Optimized Real-Time Resilient Structural Segmentation via Kolmogorov-Arnold Enhanced Spatial Attention Networks

2025-07-16

AI Summary

To address the trade-off between segmentation accuracy and computational efficiency in structural defect detection for civil infrastructure, this paper proposes a lightweight and efficient segmentation network. The architecture employs a depthwise separable convolution-based backbone, integrates an adaptive Kolmogorov–Arnold network (TiKAN) to enhance nonlinear modeling capability, and introduces a spatial-channel协同 multi-scale attention fusion mechanism to strengthen feature discriminability while significantly reducing model complexity. Experimental results demonstrate that, compared to benchmark methods including U-Net and SA-UNet, the proposed approach achieves a 91% reduction in both parameter count and FLOPs, triples inference speed, and attains an F1-score of 0.771 and a mean Intersection-over-Union (mIoU) of 0.677. These improvements establish a viable solution for high-accuracy, low-overhead real-time structural defect detection in field applications.

Technology Category

• Computer Vision: Segmentation
• Machine Learning: Hardware-aware ML
• Data Mining & Knowledge Management: Anomaly/Outlier Detection

Application Category

• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Web Mining and Content Analysis: Bridging structured and unstructured data

Abstract

Automated structural defect segmentation in civil infrastructure faces a critical challenge: achieving high accuracy while maintaining computational efficiency for real-time deployment. This paper presents FORTRESS (Function-composition Optimized Real-Time Resilient Structural Segmentation), a new architecture that balances accuracy and speed by using a special method that combines depthwise separable convolutions with adaptive Kolmogorov-Arnold Network integration. FORTRESS incorporates three key innovations: a systematic depthwise separable convolution framework achieving a 3.6x parameter reduction per layer, adaptive TiKAN integration that selectively applies function composition transformations only when computationally beneficial, and multi-scale attention fusion combining spatial, channel, and KAN-enhanced features across decoder levels. The architecture achieves remarkable efficiency gains with 91% parameter reduction (31M to 2.9M), 91% computational complexity reduction (13.7 to 1.17 GFLOPs), and 3x inference speed improvement while delivering superior segmentation performance. Evaluation on benchmark infrastructure datasets demonstrates state-of-the-art results with an F1- score of 0.771 and a mean IoU of 0.677, significantly outperforming existing methods including U-Net, SA-UNet, and U- KAN. The dual optimization strategy proves essential for optimal performance, establishing FORTRESS as a robust solution for practical structural defect segmentation in resource-constrained environments where both accuracy and computational efficiency are paramount. Comprehensive architectural specifications are provided in the Supplemental Material. Source code is available at URL: https://github.com/faeyelab/fortress-paper-code.

Problem

• Achieving high accuracy in real-time structural defect segmentation
• Balancing computational efficiency with segmentation performance
• Reducing parameters and complexity for resource-constrained environments

Innovation

• Depthwise separable convolution reduces parameters
• Adaptive TiKAN integration optimizes function composition
• Multi-scale attention fusion enhances feature combination