ABSTRACT
The development of computational intelligence techniques plays a necessary part in human daily life. Modern computer hardware and digital image processing techniques created an explosion of digital images over wide range of scientific and engineering applications, such as surveillance, image retrieval, speech processing, digital television, etc. The goal of a visual-tracking algorithm is to segment a detected object in video and observe its motion in the consecutive frames then to analyze the object's tracks to understand its behavior. The conventional image processing techniques often face great difficulties in visual tracking because of sudden transformations due to gestures, appearance, non-rigid structures, occlusion, and camera motion. Moreover, it is limited in its ability to process natural data in its raw form and real time because of the huge amount of information availability in the video. The latest advancements in deep learning added a massive boost to the computer vision field. Deep learning networks implement automatic feature extraction without human intervention and they deal with massive amount of data. They make complex decision making easier which facilitates applying video processing to make things smarter and reliable. In this chapter, the contemporary progress on object detection and visual-tracking styles has been analyzed. Research problems and possible solutions have also been explored.
