Method and Apparatus for Real-time Target Recognition Type Tracking

摘要

A real-time targeted target tracking method and apparatus is presented. The real-time targeted target tracking method proposed in the present invention generates a model for target detection using an initial target location stored in the learning data generation module, and detects and tracks target candidates based on the generated model, dip Generating target data to be used for deep learning using the target learning data generated from the initial target position and target tracking result through the learning learning data generation module, a model for the target to be used for deep learning through the deep learning target model generation module Learning, generating a model, recognizing the type of deep learning targets for individual target detection results through the deep learning target recognition module, determining an optimal target candidate using the recognized deep learning target, and subsequent deep learning Generate training data to determine the optimal target candidate to participate in learning data generation. And reflecting the module.