HIGH PERFORMANCE TARGET TRACKING DEVICE USING SUPER RESOLUTION CONVOLUTIONAL NEURAL NETWORK

摘要

A high-performance target tracking device using a super-resolution convolution neural network (SRCNN) is disclosed. A target tracking method performed by a target tracking apparatus according to an embodiment includes: receiving a target area from image information including background information having a complexity greater than or equal to a preset reference; acquiring a learning result by learning the received target region on a deep learning model generated for target tracking; detecting a target using a target model configured for target detection using the target feature extracted as the obtained learning resu and generating a target region for image information including background information having a complexity greater than or equal to a preset reference based on the detected target.