Road crack detection method based on improved FCN

Abstract

The invention relates to the technical field of computer vision and pattern recognition, in particular to a road crack detection method based on an improved FCN, which can improve the accuracy of a model on the basis of the original FCN, and adopts a pre-training method, so that computing resources are saved, and the speed is increased. The method comprises the following steps: (1) acquiring and preprocessing road crack data; (2) sending the image data, including an original image and a labeled image, into an improved FCN neural network model for training; (3) training an improved FCN neural network model according to the acquired road crack image data set; (4) in order to better evaluate the performance of the model, evaluating by adopting two indexes of total pixel accuracy and an average cross-parallel ratio; and (5) using binary classification cross entropy (BCE) as a loss function in the training process.

Description

technical field [0001] The invention relates to the technical field of computer vision and pattern recognition, in particular to a road crack detection method based on improved FCN. Background technique [0002] One of the initial manifestations of pavement disease is road cracks, and road cracks have always been an important task in road surface management. It is particularly important to find and repair road cracks in time. Road cracks pose a threat to driving safety. The traditional manual method not only fails to meet the needs of development, but also often has subjectivity and insufficient accuracy. With the development of deep learning technology, the method of neural network semantic segmentation can extract road cracks very well. [0003] FCN is a fully convolutional neural network, which can be used as a semantic segmentation neural network model for the detection of road cracks, but because FCN often does not achieve good results when facing small and long target...

AI Technical Summary

Problems solved by technology

[0003] FCN is a fully convolutional neural network, which can be used as a semantic segmentation neural network model for the detection of road cracks. However, because FCN often does not achieve good results when facing small and long targets such as cracks, its The reason is that the FCN neural network model adopts a larger step size in the crack information extraction process of convolution and in the deconvolution process, which will cause a lot of crack information to be lost in the final prediction map. Due to the constraints of computer performance, the huge neural network model will consume a lot of time during the training process, and the results cannot be obtained quickly

Images