Research On Detection Algorithm Of Rail Fastener Based On Convolutional Neural Network

Railway transportation is the artery of China’s national economy and plays an important role in the development of today’s society.Due to the late start of China’s railway security inspection technology,the current railway security inspection tasks mainly rely on manual inspection,but the manual inspection efficiency is low,and a lot of manpower and material resources are needed.With the development of the railway network and the increasement of labor costs,an automatic inspection equipment is urgently needed to achieve efficient inspection,reduce labor costs,and improve the speed and the accuracy of detection.This paper focuses on the computer vision technology in the field of artificial intelligence and image detection method,applies it to the field of rail fastener detection.In this paper,the deep convolutional neural network is used to construct the rail fastener detection model,and the deep learning model is optimized for the rail fastener detection task,realizing high-performance real-time detection and significantly improving the detection efficiency of the rail fastener.In this paper,the image data set of rail fastener detection is established,which includes four kinds of rail fastener types with a total of 4,000 sheets of rail fasteners.The region proposal network is used to generate the region of interest,the convolutional neural network is used to extract features,and the classifier is integrated into the detection network.The fastener image detection network is eatablished based on the Caffe deep learning framework and the Faster R-CNN deep learning detection framework.The accuracy of the model is improved by network replacement,data enhancement and online hard sample mining.In the final deployment environment of TITAN X graphics card,the model accuracy reached 99%and the speed reached 50FPS.