Visual Tracking of Suspicious Entities through Unmanned Aerial Vehicles in Conflicted Areas, Structures and Urban Regions

In response to security and surveillance needs, unmanned aerial vehicles are being employed to carry out monitoring and target identification in conflict scenarios, such as areas dominated by criminal factions, structures under siege by terrorists, and zones of armed conflict. And having the motivation to contribute with solutions to solve these needs, our research presents a tracking application that integrates object detection with a region-based convolutional neural network as a detector of targets of interest and the discriminative correlation filter with Channel and Reliability Spatial as a tracking algorithm for the proposed method. This approach has the objective and the motivation to help the preventive actions of security systems used in conflicted areas, in structures, and in urban regions. The generated model results show an average accuracy of 99.8% for the object tracker when applied to the video sequences in the image dataset.