Software Design Of Duty Qfficers' Fatigue And Departure Detection System Based On Video Analysis

The Armed Police Force is an important part of the Chinese armed forces and is responsible for the important care and security tasks of the country.It has a prominent role and a heavy responsibility.In the duty process of the Armed Police on-duty personnel,their status will directly affect the normal operation of the Armed Police Force.Therefore,the police service surveillance system will be deployed on a large scale.However,the traditional police service surveillance system uses manual monitoring,resulting in a huge waste of human resources.Therefore,the development of an automatic monitoring system for alternative manpower monitoring has important engineering practical value.This thesis develops a system software for fatigue and departure detection based on video analysis.The software mainly consists of server and client.The network camera is connected to the Ethernet to obtain video data.The system has developed a flexible communication mechanism for data exchange between modules,and has designed a business framework that includes modules for data reception,data decoding,anomaly detection,and alarm management.Among them,the detection module directly affects the performance of the system and consists of two parts:Fatigue detection trains the detection model through the combination of MTCNN and"Adaboost+Haar-like features" to achieve the requirement of fatigue detection.The departure detection uses the SSD to train pedestrian detection model,and continuously replenish and optimize the samples to meet the requirements for departure detection.The software adopts modularization and low coupling design,making the software more scalable and maintainable.After testing,the fatigue and departure detection system software has high accuracy in various environments,including scenes with large differences in light such as day/night,scenes with single/multi-target and scenes where the shape of the subject was different(front/side,wearing glasses,etc.).The accuracy of the fatigue detection achieved 95.9%and 92.9%,and the departure detection achieved 98.3%and 95.6%during daytime and nighttime.Therefore,the system can meet the actual needs of the automatic monitoring of on-duty personnel's status,and has been put into use in actual scenarios.