Association Technology With Baggage And Passengers In The Railway Passenger Stations Based On Target Detection

In the densely populated and complex railway passenger stations,the loss of passengers’ baggage occurs from time to time,which seriously affects the travel of passengers.In this thesis,combined with the technologies of target detection,case segmentation and pedestrian re-identification,passengers and baggage association technology based on target detection is proposed in railway passenger stations to reduce the situation of passengers taking their baggage by mistake,so as to ensure the property safety of passengers.In order to judge the relationship between passengers and baggage,this thesis proposes a passengers and baggage association technology based on target detection to associate passengers and their baggage in the railway passenger stations.Firstly,the YOLACT model is used to detect the video frames to obtain the location information and category of passengers and baggage.The experimental results show that the mean accuracy(m AP)of detection using YOLACT in different situations exceeds 85,which can meet the detection requirements in station scenarios.Secondly,an association relationship determination method based on the relative position information of passengers and baggage is proposed to determine the relationship between passengers and baggage in the image.In order to adapt to the diversity of practical application scenarios,this thesis proposes two relationship determination methods.Among them,the association relationship determination method based on the overlap area of passengers and baggage realizes the determination of the association relationship by calculating the proportion of the overlap area of passengers and baggage in the baggage area;The association relationship determination method based on the location information of the center point realizes the association by judging the distance between the passenger and the baggage center point.The experimental results show that the accuracy rate of the association relationship determination method based on the overlapping area of passengers and baggage is about 78.60%,and the accuracy rate of the association relationship determination method based on the location information of the center point is about 80.73%.Finally,the association relation selection based on the relation matrix is used to comprehensively process the image judgment results,and the selection is made according to the frequency of the passengers and baggage being judged to have an association relationship,so as to obtain the final passengers and baggage association information.The experimental results show that the association relation selection method based on relation matrix can reduce the impact of the change of the relative position of passengers’ baggage in the video,and effectively improve the accuracy of association relation determination.In view of the baggage loss in the process of passenger travel,this thesis proposes the process of passengers and baggage association and identification in the railway passenger stations in combination with the passengers and baggage association technology,in order to ensure the property safety of passengers in the process of travel.Firstly,the images of passengers when they enter the station or check their baggage are processed,and the passengers and baggage association technology based on target detection is used to obtain the associated information of passengers and baggage.When the passenger needs to pick up the baggage,the pedestrian re-identification technology is used to find the target passenger in the passenger list,and then judge whether the passenger takes the correct baggage according to the previously obtained related information.Collecting station videos to simulate the process of passengers and baggage association and identification in the railway passenger stations.The experimental results show that in the process of passengers and baggage association,the accuracy of the association relationship determination method based on the overlapping area of passengers and baggage is about 80.25%,and the accuracy of the association relationship determination method based on the location information of the center point is about 83.76%.The accuracy of the two methods in practical application is roughly the same as the test results.In addition,in the process of identification,the accuracy rate of the passenger’s baggage retrieval result is about 75.13%,which can better identify and judge the passenger’s baggage relationship,and provide support and help to prevent the loss of passenger’s baggage.