Self-Learning Vehicle Detection and Tracking from UAVs

Vehicle detection and tracking with unmanned aerial vehicles (UAVs) find increasingly widespread applications in both military and civilian domains. In this paper, a method for vision-based multiple vehicle detection by using the proposed self –learning tracking and detection (SLTD) method has been proposed. The method used Features from Accelerated Segment Test (FAST) and Histograms of Oriented Gradients (HoG) for vehicle detection. Based on the detection results, a Forward and Backward Tracing (FBT) mechanism has been employed in the new self-learning tracking algorithm based on Scalar Invariant Feature Transform (SIFT) feature. The main aim of this research is to improve the accuracy of the detection and tracking system, where the detector relies on the features of a pre-trained model with no connection with the current detection or tracking. The main contribution of this paper is that the proposed system can detect and track multiple vehicles with a self-learning process leading to increase the tracking and detection accuracy. UAV videos captured in different situations have been used to evaluate the proposed algorithm. The results demonstrated that the accuracy can be improved by using the proposed method.