Small object detection in UAV image based on slicing aided module

Abstract

Unmanned aerial vehicles (UAVs) play an increasingly important role in today’s life, and play a pivotal role in many fields such as remote sensing, command, and agriculture. UAVs have the advantages of high-altitude hovering, compactness and convenience, and are one of the most ideal platforms for taking photos in the air. Since the drone can shoot at high altitude, the target object in the image is often small in size, which is easy to cause it to be blurred and difficult to distinguish. With the development of artificial intelligence technology, the application of computer vision technology to build deep learning models can automatically identify large-scale and small targets from UAV images. Although this field has been concerned by many scholars, how to deploy small target detection on UAVs and apply it to command and control remains to be studied. To this end, this study proposes a small object detection method based on a slicing aided module to automatically detect objects from UAV images. The proposed method can be widely used in scenarios such as on-site command and control, which is of great significance for the further wide application of UAVs.