Research On Object Determination And Tracking Method Of Shooting Confrontation Training System

As a competitive sport with a strong focus on professionalism and high technical competency,shooting plays a critical role in the field of both military training and competition.Currently,the most common shooting training methods use target shooting on static fixed targets,but these methods are incapable of providing training instructors with a genuine combat situation,and some risks exist.Consequently,there is significant research significance and practical value associated with researching how to apply dynamic shooting training through man-machine confrontation.This paper's focus is on deep learning and intelligent decision control techniques.It is intended to help with the design of a shooting confrontation training system and to explore how it makes decisions and follows targets.Research topics and findings include:(1)Proposed the shooting confrontation training system's architectural design.This article creates a training system that combines object detection and behavior tree algorithm to replicate actual combat conditions on the battlefield.This entire system,which consists of four modules: image acquisition,object detection,bus communication,and object tracking,is part of a completely self-contained package that is able to be detached from the ground.The object detection and determination tracking of person targets is comprehensively implemented.(2)Established object detection and object tracking algorithms for the shooting confrontation training system.Combined with the deep learning method,this scheme design an SSD(Single Shot Multi Box Detector)object detection algorithm based on the Inception network structure.By collecting the original image data of the person targets in standing,squatting,and crawling positions and training the network model,this scheme enhances the feature extraction capability of the network model and accomplishes highprecision detection of person targets.Designed the object tracking algorithm,calculating the object tracking angles for determining the rotation angles required by the cloudcomputing to track the target,analyzing the cloud-computing control logic,and realizing the real-time control and object tracking of the cloud-computing system.(3)Forced a method of determining the object's identity during shooting confrontations.This algorithm sets decision-making conditions,such as threshold parameters and quantitative indicators,as well as an object attack sequence,for a multiobjective decision algorithm and creates a system with both a Behavior Tree method and decision rules for it.Verify the algorithm's decision-making ability using the object determination tests.(4)Constructed the shooting engagement training system's test platform.Selecting and designing the system's subsystem functions to allow cooperative application and realtime information interaction is crucial for proper image acquisition and object tracking.Constructed the shooting conflict training system's long-distance object tracking shot test and the short-distance object decision-making shot test,and demonstrated that this system was effective and suitable for engineering use.