Automated Pavement Crack Segmentation Using U-Net-Based Convolutional Neural Network

Abstract

Automated pavement crack image segmentation is challenging because of inherent irregular patterns, lighting conditions, and noise in images. Conventional approaches require a substantial amount of feature engineering to differentiate crack regions from non-affected regions. In this paper, we propose a deep learning technique based on a convolutional neural network to perform segmentation tasks on pavement crack images. Our approach requires minimal feature engineering compared to other machine learning techniques. We propose a U-Net-based network architecture in which we replace the encoder with a pretrained ResNet-34 neural network. We use a "one-cycle" training schedule based on cyclical learning rates to speed up the convergence. Our method achieves an F1 score of 96% on the CFD dataset and 73% on the Crack500 dataset, outperforming other algorithms tested on these datasets. We perform ablation studies on various techniques that helped us get marginal performance boosts, i.e., the addition of spatial and channel squeeze and excitation (SCSE) modules, training with gradually increasing image sizes, and training various neural network layers with different learning rates. INDEX TERMS Convolutional neural network, deep learning, fully convolutional network, pavement crack segmentation, U-Net.

Differentiating Small-Cell Lung Cancer From Non-Small-Cell Lung Cancer Brain Metastases Based on MRI Using Efficientnet and Transfer Learning Approach

“…Due to the relatively small data size that was available for this study, the network was trained by a pre-trained model based on EfficientNet-b0 pre-trained weights that had been trained on ImageNet data set as previously described. 17 , 18 …”

Automatic segmentation, classification, and follow‐up of optic pathway gliomas using deep learning and fuzzy c‐means clustering based on MRI

“…Following optimization, the following training parameters were used: batch size = 4, for the network training adaptive moment estimator (ADAM) stochastic gradient based optimizer 19 was used with initial learning rate of 1e −5 , and a total of 125 epochs. The network training was initiated using a pre‐trained model (ResNet‐34‐model weights, trained on ImageNet dataset, 100 epochs), followed by fine‐tuning of the model by unfreezing the weights (additional 25 epochs), as previously described 20,21 . The dice score was used to evaluate the model during training.…”

“…The network training was initiated using a pre-trained model (ResNet-34-model weights, trained on ImageNet dataset, 100 epochs), followed by fine-tuning of the model by unfreezing the weights (additional 25 epochs), as previously described. 20,21 The dice score was used to evaluate the model during training. DL model evaluation: Segmentation results of the entire tumor area were evaluated using fivefold cross-validation, while ensuring that images belonging to a given patient would be allocated only within one dataset (training, validation, and testing).…”

Oil Identification on Beaches Using Deep Learning Techniques

“…The main advantage of ResNet family compared to others CNNs is its expertise on computer vision tasks, possible due to the capability of creating shortcut connections which allow the simplification of information between the layers, optimizing the training [14]. Therefore, the ResNet-34 plays an important role as in Lau et al [15], that proposes a model instructed to identify cracks in sidewalks. That's because the irregular standards and lighting conditions make the process difficult, which is mitigated by ResNet-34 functions.…”