Multiple Flying Object Detection using AlexNet Architecture for Aerial Surveillance Applications
Abhishek Bajpai, Vaibhav Srivastava, Shruti Yadav, Yash Sharma · 2023
In the past few years, drones have been used in military applications. Flying objects, such as mini UAVs, are increasingly being used maliciously by terrorists, criminals, and smugglers. These more technologically advanced and affordable devices have a high probability of assault, occur frequently, and have the potential for disastrous impacts. The risk of bird strikes always exists, because birds and aircraft occupy the same space at low elevations. Although there are restrictions on flying drones close to airports and designated airspaces to prevent them from invading each other’s spaces, there are still a number of difficulties. To overcome such challenges, the detection and tracking of various flying objects are key tasks. In this study, Two Convolutional Neural Network(CNN) models were employed, that is (ResNet101 and AlexNet), and the trade-off between these two architectures gave us a highly efficient and accurate system with a hopping training accuracy of 93.68% that can not only easily track and detect but also classify it with high accuracy.