NEFELI: A computer vision software for aircraft detection and tracking implemented on an edge Graphic Processing Unit

<p>Efficient detection and accurate collision estimation for non-cooperative aerial vehicles are crucial for the realization of fully autonomous aircraft. This paper</p>\n<p>introduces NEFELI, a machine learning software utilizing optical sensors to detect, and track non-cooperative aerial vehicles. NEFELI’s detector employs an enhanced YOLOv5 large model, strengthened with a sliding window postprocessing step to enhance detection capabilities for distant, diminutive objects. Furthermore, NEFELI introduces several innovations in its tracking component.</p>\n<p>The tracking system adopts the Strong-SORT model, enhanced with a global motion compensation technique to accommodate the movement of the host aircraft’s</p>\n<p>camera. A key advancement lies in the creation and utilization of the first large-scale re-identification dataset of aerial objects. This dataset is used to train the deep learning component of the tracking module. </p>