🤖 AI Summary
Deploying crack segmentation for structural health monitoring (SHM) on resource-constrained edge microcontrollers—limited in memory, computational capacity, and energy—is highly challenging.
Method: We propose the first TinyML-optimized U-Net system for edge SHM, jointly optimizing model size, memory footprint, and latency via filter reduction, depth compression, depthwise separable convolutions (DWConv2D), and structured pruning.
Contribution/Results: Our lightweight architecture achieves 4.2% of the original U-Net’s parameter count, reduces RAM usage by 83%, and accelerates inference by 5.6×—while preserving crack localization and dimensional estimation accuracy. It fits within tight Flash/RAM budgets and enables long-term operation on battery-powered or even energy-harvesting edge devices. This work establishes a reproducible technical pathway and empirical benchmark for deploying lightweight semantic segmentation in ultra-low-power embedded SHM systems.
📝 Abstract
Crack segmentation can play a critical role in Structural Health Monitoring (SHM) by enabling accurate identification of crack size and location, which allows to monitor structural damages over time. However, deploying deep learning models for crack segmentation on resource-constrained microcontrollers presents significant challenges due to limited memory, computational power, and energy resources. To address these challenges, this study explores lightweight U-Net architectures tailored for TinyML applications, focusing on three optimization strategies: filter number reduction, network depth reduction, and the use of Depthwise Separable Convolutions (DWConv2D). Our results demonstrate that reducing convolution kernels and network depth significantly reduces RAM and Flash requirement, and inference times, albeit with some accuracy trade-offs. Specifically, by reducing the filer number to 25%, the network depth to four blocks, and utilizing depthwise convolutions, a good compromise between segmentation performance and resource consumption is achieved. This makes the network particularly suitable for low-power TinyML applications. This study not only advances TinyML-based crack segmentation but also provides the possibility for energy-autonomous edge SHM systems.