Efficient Application of AI for Target Tracking and Monitoring in Airborne Images:

With evolution of computing technology, its capability of solving complex problems faced in real life scenario have certainly escalated. One such problem of extraction and counting of building footprints using aerial images obtained from drone is looked upon in this paper.A typical image obtained from drone has many characteristic features like the spatial resolution i.e. the area that a single pixel of the image covers, pixel resolution and many more, and provides users with a 3 channel(RGB) image of a certain region which helps in assessing the landscape distribution, vegetation cover urbanization and monitoring the area. With so many efficient and intriguing applications, drone images have seen a definite upsurge in recent years. However, all such applications are painstakingly tedious and certainly require automation. Researchers are constantly working towards solving this problem and providing an optimal solution economically and computationally feasible.This paper proposes to solve this problem by carrying out basic image processing steps enhanced using Deep Learning by efficiently harnessing the statistical features of image and applying adaptive thresholding to semantically segment and count building footprints, thus deriving a holistic algorithm that is capable of handling image with different spatial or pixel resolution.