Data fusion for unsupervised video object detection, tracking and geo-positioning

In this work we describe a system and propose a novel algorithm for moving object detection and tracking based on video feed. Apart of many well-known algorithms, it performs detection in unsupervised style, using velocity criteria for the objects detection. The algorithm utilises data from a single camera and Inertial Measurement Unit (IMU) sensors and performs fusion of video and sensory data captured from the UAV. The algorithm includes object tracking and detection, augmented by object geographical co-ordinates estimation. The algorithm can be generalised for any particular video sensor and is not restricted to any specific applications. For object tracking, Bayesian filter scheme combined with approximate inference is utilised. Object localisation in real-world co-ordinates is based on the tracking results and IMU sensor measurements.