Research On Key Behavior Of Air Baggage Self-service Check-in

To improve the service experience brought to users in human-computer interaction effectively,we need to recognize and detect some key behavioral information in the air passenger self-service baggage check-in process.The difficulty of real-time online detection of passenger behavior lies to ensuring both recognition accuracy and recognition speed.To this end,a high-speed behavior recognition algorithm and online behavior detection framework is studied to achieve effective recognition and detection of key behaviors in the self-service check-in operating system.Firstly,considering the necessity of behavior and the objects involved,analyzing the actual interaction process between passengers and the self-service baggage check system,based on these,we confirmed the need to identify key behaviors,and collect samples in the airport environment and laboratory environment respectively,eventually,established two datasets for the self-service check-in key behavior identification task.Secondly,a deep dynamic time-domain segmentation network is proposed based on the temporal segmentation network behavior recognition algorithm,which improves the recognition speed while maintaining high recognition accuracy.Again,the recognition accuracy and recognition speed of the deep dynamic temporal segment networks are verified on the public dataset,and the results show that the algorithm speed is greatly improved compared with the classical high-speed behavior recognition algorithm,while maintaining a high recognition accuracy.Tested on two self-constructed datasets,the results show that the newly proposed algorithm can better solve the task oriented to self-service consignment critical behavior recognition.Finally,a behavior detection framework based on a deep dynamic temporal segment networks is designed and tested on a self-built dataset.The results show that the newly proposed algorithm has high detection accuracy and good recognition speed on multiple hardware platforms,meeting the accuracy requirements and real-time requirements of online detection.