DEEP LEARNING-BASED CRACK, LOCATION AND AREA IDENTIFICATION FOR A PIPELINE BY THE CONVOLUTIONAL NEURAL NETWORK BASED ON CRACK CONTOUR NETWORK METHOD

In this work, the Convolutional Neural Network (CNN) algorithm is introduced in pipeline surface cracks monitoring-based image processing method for improving the efficiency and accuracy of crack type, location, and area identification.The method is used to extract the cracks area called the CNN based on crack contour network (CCN-CNNs) method from locate and extract the crack shape.CCN-CNNs is provides the accuracy rate (P%), recall rate (R%), and F-score (F%) index to assess the algorithm in the problem while identifying the cracks, and then according to the maximum F-score, we computes the crack corresponding contour area.In this work the pipeline crake images datasets are provided using an inspection drone with high definition camera.To the best of the authors' knowledge, the methodology presented in this paper for pipeline crack identification is an original contribution to the literature.This work introduces an efficient approach that also significantly reduces the time for crack type, location, and area identification of pipelines, the accuracy rate (P%), recall rate (R%), and Fscore (F%) are recorded 91.8% ,86.1%, and 84.6% respectively.