Railway intrusion foreign matter unmanned aerial vehicle detection method, device and system based on deep learning

Abstract

The invention discloses a deep learning-based railway intrusion foreign matter unmanned aerial vehicle detection method and system. The method comprises the steps of S1, acquiring video image data, collected by an unmanned aerial vehicle, along a to-be-detected railway in real time; s2, detecting whether a foreign matter exists in the video image data by using a deep learning-based target detection model, and when the existence of the target foreign matter is detected, turning to execute a step S3; s3, the railway clearance in the target video frame with the target foreign matter is extracted, the railway clearance is the clearance where the foreign matter is not allowed to invade on the railway, and whether the target foreign matter invades the railway clearance or not is judged according to the detected position state of the target foreign matter. The method has the advantages of being simple in implementation method, high in detection precision and efficiency, wide in detection range, flexible and the like.

Description

technical field [0001] The present invention relates to the technical field of rail transit detection, in particular to a deep learning-based method, device and system for detecting railway intrusion and foreign object unmanned aerial vehicles. Background technique [0002] The intrusion of foreign matter into the railway boundary will cause great harm to the safety of railway traffic, which may delay the arrival time of the train at the station, and may also cause economic losses. According to statistics, the faults of the railway system are caused by the equipment itself, which usually only accounts for 20%, and most of the faults are caused by external factors such as foreign matter invasion. With the rapid development of railway construction, the speed of trains continues to increase, and the requirements for train operation safety are also constantly increasing, while the hidden dangers of environmental safety along the railway line are also becoming more and more promi...

AI Technical Summary

Problems solved by technology

[0006] For foreign objects on railways, manual inspection is usually used at present, but manual inspection is inefficient, expensive, and prone to missed inspections, such as small-sized stones, which are not easy to be found, relying on manual inspections It is very difficult to realize real-time and accurate investigation of the entire railway line

In order to solve the above problems, one solution is to install a camera next to the railway line that needs to be monitored, collect images through the camera, design an image recognition algorithm in advance based on the target characteristics that need to be identified (such as vehicles, human bodies, etc.), and collect images in real time. Image processing, using the designed image recognition algorithm to identify whether there is a target in the image, but this type of solution can only achieve small-scale monitoring at a specific location, and can only identify specific types of targets, and foreign object intrusion events occur on the railway It is sudden and random, and the time and place of its occurrence are often difficult to predict, and the total amount of railway mileage is large, and there may be foreign object incidents along the entire railway line. The above-mentioned image recognition method cannot realize the detection of foreign objects along the entire railway line. Foreign object monitoring, and as shown in Table 1, there are many types of foreign objects with different morphological characteristics. To realize the detection of various foreign objects, the design of image recognition algorithms is very difficult. There are many types of foreign objects to be identified, which will also affect the efficiency of recognition. , unable to meet the timeliness requirements for foreign object detection during high-speed train operation

Images