Abstract:Despite recent breakthroughs in deep learning and computer vision fields, the pixel-wise identification of tiny objects in high-resolution images with complex disturbances remains challenging. This study proposes a modified U-net for tiny crack segmentation in real-world steel-box-girder bridges. The modified U-net adopts the common U-net framework and a novel Self-Attention-Self-Adaption (SASA) neuron as the fundamental computing element. The Self-Attention module applies softmax and gate operations to obtain the attention vector. It enables the neuron to focus on the most significant receptive fields when processing large-scale feature maps. The Self-Adaption module consists of a multiplayer perceptron subnet and achieves deeper feature extraction inside a single neuron. For data augmentation, a grid-based crack random elastic deformation (CRED) algorithm is designed to enrich the diversities and irregular shapes of distributed cracks. Grid-based uniform control nodes are first set on both input images and binary labels, random offsets are then employed on these control nodes, and bilinear interpolation is performed for the rest pixels. The proposed SASA neuron and CRED algorithm are simultaneously deployed to train the modified U-net. 200 raw images with a high resolution of 4928 x 3264 are collected, 160 for training and the rest 40 for the test. 512 x 512 patches are generated from the original images by a sliding window with an overlap of 256 as inputs. Results show that the average IoU between the recognized and ground-truth cracks reaches 0.409, which is 29.8% higher than the regular U-net. A five-fold cross-validation study is performed to verify that the proposed method is robust to different training and test images. Ablation experiments further demonstrate the effectiveness of the proposed SASA neuron and CRED algorithm. Promotions of the average IoU individually utilizing the SASA and CRED modules add up to the final promotion of the full model, indicating that the SASA and CRED modules contribute to the different stages of model and data in the training process.

Microstructure Recognition of Steels by Machine Learning Based on Visual Attention Mechanism

Quantitative Analysis of Metallographic Image Using Attention-Aware Deep Neural Networks

Microstructural segmentation using a union of attention guided U-Net models with different color transformed images

Complex-Phase Steel Microstructure Segmentation Using UNet: Analysis across Different Magnifications and Steel Types

Feature Extraction and Microstructural Classification of Hot Stamping Ultra-High Strength Steel by Machine Learning

Microstructure Image Segmentation of 23crni3mo Steel Carburized Layer Based on a Deep Neural Network

HEU-Net: Hybrid Attention Residual Block-Based Network with External Skip Connections for Metal Corrosion Semantic Segmentation

Spatiotemporal Prediction and Characterization of Microstructure Evolution During In-Situ Heat Treatment Using a Deep Learning Algorithm

Study on the Recognition of Metallurgical Graphs Based on Deep Learning

Automated Segmentation of Large Image Datasets using Artificial Intelligence for Microstructure Characterisation, Damage Analysis and High-Throughput Modelling Input

Advanced Steel Microstructural Classification by Deep Learning Methods

Development and application of automatic identification methods based on deep learning for oxide scale structures of iron and steel materials

Segmentation method of U-net sheet metal engineering drawing based on CBAM attention mechanism

Intelligent Detection of Steel Defects Based on Improved Split Attention Networks

Microstructure quality control of steels using deep learning

High throughput quantitative metallography for complex microstructures using deep learning: A case study in ultrahigh carbon steel

A modified U-net for crack segmentation by Self-Attention-Self-Adaption neuron and random elastic deformation

Intelligent Classification of Metallographic Based on Improved Deep Residual Efficiency Networks

Exploring the microstructure manifold: Image texture representations applied to ultrahigh carbon steel microstructures

A Multiple Feature-Maps Interaction Pyramid Network for Defect Detection of Steel Surface

Dual attention deep learning network for automatic steel surface defect segmentation