Locomotive driver behavior recognition method, device and system

Abstract

The present invention provides a locomotive driver behavior recognition method, a device and a system. The locomotive driver behavior recognition method comprises the steps of preliminarily defining a to-be-trained image set containing multiple types of driver behaviors; based on the neural network algorithm, subjecting the to-be-trained image set to model training so as to generate a behavior training model; acquiring the real-time monitoring image of a locomotive cab, inputting the real-time monitoring image in to the behavior training model, and recognizing the type of a driver behavior. According to the technical scheme of the invention, based on the deep learning algorithm, several types of driver daily operations are automatically recognized. Meanwhile, images are automatically retrieved, so that the heavy manual work is avoided. The analysis efficiency and the analysis accuracy are improved.

Description

technical field [0001] The invention relates to rail transit locomotive vehicle safety monitoring technology, in particular to a locomotive driver behavior recognition method, device and system in a driving cab. Background technique [0002] Traffic safety has now become a social issue that is generally concerned, and it is imminent to strengthen the control of railway traffic safety. At present, the video surveillance equipment of the locomotive on-board safety protection system (referred to as 6A system) is installed on the locomotive. At present, the video surveillance equipment The ground analysis task of video data usually adopts the method of manual browsing and viewing. On the one hand, analyzing the video recordings of all locomotives requires a large number of sufficient ground video data analysis personnel and equipment, and the analysis efficiency is low. Especially in the case of specific event search, the analysis The workload is huge and easily missed. On the ...

AI Technical Summary

Problems solved by technology

For example, when the locomotive is running, the vibration amplitude is relatively large, and the image is blurred, which affects the stability of the video image. The illumination changes during the train operation are relatively large, which greatly affects the accuracy of recognition. The recognition objects in the locomotive environment are quite different, which will also affect video recognition. adaptability, etc.

In addition, when the train enters and exits the station, there may be strong light conditions. At this time, the effect of video analysis and recognition will be reduced due to the influence of strong light. When the light returns to normal, the recognition effect will also improve. How to reduce the impact of illumination changes and improve the robustness of the algorithm is one of the technical difficulties faced by the industry

Images