Date Approved


Embargo Period


Document Type


Degree Name

M.S. Electrical and Computer Engineering


Electrical and Computer Engineering


Henry M. Rowan College of Engineering


Ghulam Rasool, Ph.D.

Committee Member 1

Nidhal C. Bouaynaya, Ph.D.

Committee Member 2

Charles C. Johnson


Computer Vision, Deep Learning, Machine Learning


Artificial intelligence; Computer simulation


Aviation Safety and Security | Electrical and Computer Engineering


The updated information about the location and type of rotorcraft landing sites is an essential asset for the Federal Aviation Administration (FAA) and the Department of Transportation (DOT). However, acquiring, verifying, and regularly updating information about landing sites is not straightforward. The lack of current and correct information about landing sites is a risk factor in several rotorcraft accidents and incidents. The current FAA database of rotorcraft landing sites contains inaccurate and missing entries due to the manual updating process. There is a need for an accurate and automated validation tool to identify landing sites from satellite imagery. This thesis proposes an AI-based approach to scan large areas using satellite imagery, identify potential landing sites, and validate the FAA's current database. The proposed method uses the object detection technique, one of the well-known computer vision methods used to identify objects of interest from image data. Objection detection techniques are based on the famous convolutional neural networks (CNN) and have achieved state-of-the-art performance. We used FAA's 5010 database to build a satellite imagery dataset that contained manually verified landing sites, including helipads, helistops, helidecks, and helicopter runways. We explored different object detection models, including single-shot detector (SSD), you only look once (YOLO), and various flavors of mask regional CNN (R-CNN). Each model presented a unique accuracy-computational complexity trade-off. After achieving satisfactory performance, we used our selected model to search and scan satellite images downloaded from Google Earth for potential landing sites that may or may not be part of the FAA's database. The model identified 1435 new landing sites and increased FAA's current database by 46%. We also identify methods to improve our proposed model in the future.