Tracking Objects Based on Multiple Particle Filters for Multipart Combined Moving Directions Information.

Object tracking is an important procedure in the computer vision field as it estimates the position, size, and state of an object along the video's timeline. Although many algorithms were proposed with high accuracy, object tracking in diverse contexts is still a challenging problem. The paper presents some methods to track the movement of two types of objects: arbitrary objects and humans.