Railway Foreign Body Intrusion Detection Based on Faster R-CNN Network Model

Pedestrians, vehicles and other foreign bodies intruding into the railway boundaries seriously threaten the safety of pedestrians and railway traffic. In view of the shortcomings of the traditional railway foreign body detection algorithm, such as low recognition accuracy, unclear classification and the results susceptible to the external environment, a railway foreign body intrusion detection algorithm based on Faster R-CNN network model was proposed, and adaptive improvements were made to the model to meet the practical needs of railway foreign body intrusion detection. This paper proposed replacing the full connection layer with the global average pooling layer to reduce the number of parameters. An increase in the number of anchors of RPN network was proposed to improve the accuracy of the recommendations for the target area. The transfer learning was introduced to train the network and solve the problem of lack of data in the field of railway foreign body intrusion. The experimental results on the data set of railway foreign body intrusion collected by video show that the accuracy of the algorithm is 97.81% in the detection of people, vehicles and some animals. © 2020, Department of Journal of the China Railway Society. All right reserved.