Introduction to visual tracking in video sequences
Ashish Kumar · 2023
Tracking algorithms aim to predict the target state in a video stream. The tracking framework can be broadly categorized into traditional and deep learning (DL)-based architectures. Further, the conventional work can track the target by exploiting any approach that includes stochastic, deterministic, generative, or discriminative. In addition, the tracking models based on their feature extraction, which are explored under the DL domain, can be classified into deep features-based trackers or hierarchical features-based trackers. In this chapter, we discuss the various tracking problems along with the tracking challenges. Also, the limitations and benefits of the emergence of traditional tracking approaches to DL approaches are discussed, as well as elaborating on advances in the field of visual tracking and real-time applications.