Recently object tracking and detection backed by Unmanned Aerial Vehicles (UAV) have gained a lot of interest specifically in research fields like security surveillance, traffic monitoring, and search and rescue operations in a natural disaster situat...
Recently object tracking and detection backed by Unmanned Aerial Vehicles (UAV) have gained a lot of interest specifically in research fields like security surveillance, traffic monitoring, and search and rescue operations in a natural disaster situation. The key challenges in visual tracking with consumer UAVs appear in the form of continuous change in the appearance of the object of interest, target representation, target object detection, and localization in real-time. Consistent object detection mainly depends on various factors that include image noise, obstructions in the line of sight, variance in lightning conditions, posture changes, and hazy or blurriness in the image that might affect object labeling. To counter this problem, this article proposes an approach in which tracked objects in a scene is entirely described by adding contextual information, i.e., natural features, locality of the scene, and general points of interest. Each situation is described semantically by ontological statements that define the environment.