关键词:
Deep learning
visual tracking
data-invariant
data-adaptive
general components
摘要:
Recently,deep learning has achieved great success in visual tracking tasks,particularly in single-object *** paper provides a comprehensive review of state-of-the-art single-object tracking algorithms based on deep ***,we introduce basic knowledge of deep visual tracking,including fundamental concepts,existing algorithms,and previous ***,we briefly review existing deep learning methods by categorizing them into data-invariant and data-adaptive methods based on whether they can dynamically change their model parameters or ***,we conclude with the general components of deep *** this way,we systematically analyze the novelties of several recently proposed deep ***,popular datasets such as Object Tracking Benchmark(OTB)and Visual Object Tracking(VOT)are discussed,along with the performances of several deep ***,based on observations and experimental results,we discuss three different characteristics of deep trackers,i.e.,the relationships between their general components,exploration of more effective tracking frameworks,and interpretability of their motion estimation components.