Application of deep learning in computer vision for the target detection-tracking system on shoulder-training flight vehicles

Abstract

The paper presents an application of a deep learning model in computer vision for the target detection and tracking system on shoulder-training flight vehicles. The purpose of the research is to use the digital camera system as a simulated detection and tracking system similar to the features of the seeker of shoulder-launched flight vehicles, as a basis for building shoulder-training flight vehicles. To detect and track moving targets outdoors, 3D target detection ability must meet accuracy requirements; however, when traditional image processing algorithms are applied, they are not effective. To solve this problem, the article focuses on researching and applying the YOLO deep learning model to the digital camera tracking system. Using the YOLO version 8 model, the authors tested data collection, image processing and model training as well as considered the target detection and tracking ability of a digital camera system. At the same time, a newly designed shoulder-training flight vehicle - 72 millimeters with a digital camera tracking system was tested outdoors. As a result, detecting and tracking the target by the digital camera system similar to the shoulder-launched flight vehicle - 72 millimeters in a specifically limited range condition is obtained.