A comparison of convolutional object detectors for real-time drone tracking using a PTZ camera
Jihun Park, Dae Hoe Kim, Young Sook Shin, Sang-Ho Lee · 2017
As highly maneuverable drones are available at the low price, the threats that might be caused by the drone attacks has been increased. Recent object detectors have been dramatically improved in accuracy by using convolutional neural networks, and these can be utilized to identify hostile drones. In this paper, we examine state-of-the-arts convolutional object detectors for a real-time drone detection and tracking system using a Pan-Tilt-Zoom (PTZ) camera. In the drone detection and tracking system, an object detector is used to identify whether an image from the PTZ camera contains a drone, and our system generates PTZ actions to track the detected drone. To detect small size drones in real-time, an appropriate object detector should be selected. This paper compares six convolutional object detectors in the accuracy and speed.