Pixel-wise Road Pavement Defects Detection Using U-Net Deep Neural Network

Augustaukas, Rytis; Lipnickas, Arūnas

doi:10.1109/idaacs.2019.8924337

Cited by 13 publications

(8 citation statements)

References 17 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In this work, we continued our investigation on computer vision-based pavement crack segmentation by utilizing a convolutional neural network. This is an extension of our previous work [26] presented at the IDAACS'2019 conference. The mentioned article mainly focused on the classical U-Net encoder-decoder architecture depth (number of convolutional layers) and convolutional filter size dependency on the model prediction precision and computational time.…”

Section: Introductionsupporting

confidence: 57%

“…In this paper, we extended and improved our previous work [26] on pixelwise pavement crack detection by using a convolutional neural network. An investigation of road crack segmentation was scaled up by introducing additional datasets, Crack500 and GAPs384.…”

Section: Discussionmentioning

confidence: 94%

“…In addition to the original structure described in [15], we added padding to the convolutional layers to maintain the output dimension equal to the given input image. From our previous research [26], we took the best-performing solution for crack detection. It was observed that a deeper convolutional neural network tends to learn features more accurately than smaller architectures.…”

Section: Deep Neural Network Modelmentioning

confidence: 99%

“…For this mentioned reason, dropout was removed. Weight decay From our previous research [26], we took the best-performing solution for crack detection. It was observed that a deeper convolutional neural network tends to learn features more accurately than smaller architectures.…”

Section: Deep Neural Network Modelmentioning

confidence: 99%

“…For the present research, we utilized the original implementation of the residual block proposed in [35]. Additional to previous research [26], batch normalization [29] was added between the convolutional layer and its activation function (except for the last output layer with sigmoid) to increase the neural network stability. When the mean and variance are calculated using batch normalization for a small minibatch compared to the whole dataset (in this case, 4), it gives noise related to the individual iteration.…”

Section: Residual Blocksmentioning

confidence: 99%

See 4 more Smart Citations

Improved Pixel-Level Pavement-Defect Segmentation Using a Deep Autoencoder

Augustauskas

Lipnickas

2020

Sensors

View full text Add to dashboard Cite

Convolutional neural networks perform impressively in complicated computer-vision image-segmentation tasks. Vision-based systems surpass humans in speed and accuracy in quality inspection tasks. Moreover, the maintenance of big infrastructures, such as roads, bridges, or buildings, is tedious and time-demanding work. In this research, we addressed pavement-quality evaluation by pixelwise defect segmentation using a U-Net deep autoencoder. Additionally, to the original neural network architecture, we utilized residual connections, atrous spatial pyramid pooling with parallel and “Waterfall” connections, and attention gates to perform better defect extraction. The proposed neural network configurations showed a segmentation performance improvement over U-Net with no significant computational overhead. Statistical and visual performance evaluation was taken into consideration for the model comparison. Experiments were conducted on CrackForest, Crack500, GAPs384, and mixed datasets.

show abstract

Section: Introductionsupporting

confidence: 57%

Section: Discussionmentioning

confidence: 94%