Deep Learning Based Visual Tracking: A Review

Abstract

Visual tracking, a traditional computer vision task, has been a popular research field in recent decades. As a powerful features learning method, deep learning provides a new way for the realization of visual tracking with higher accuracy and performance. Many novel trackers that based on different network models, including auto-encoder (SAE), convolutional neural network (CNN), recurrent neural networks (RNN) deep reinforcement learning (DRL) and the fusion of them were proposed by researchers in the literatures. This paper presents a comprehensive survey on deep learning based visual tracking algorithms. Introduction