Is this project an undergraduate, graduate, or faculty project?
Undergraduate
individual
What campus are you from?
Daytona Beach
Authors' Class Standing
Logan Luna, Sophomore
Lead Presenter's Name
Logan Luna
Faculty Mentor Name
Richard Stansbury
Abstract
In the realm of aviation, accurate aircraft recognition and localization are of paramount importance. This study delves into the application of computer vision in conjunction with aerial maps to enhance aircraft recognition, location estimation, and real-time tracking. The methodology focuses on developing an autonomous system for identifying and localizing aircraft, along with evaluating integrated computer vision against standalone methods. A significant part of the exploration focuses on monocular depth estimate (MDE), noted for its efficacy at extended ranges, cost efficiency, and simplicity. Merging MDE with deep learning, especially Convolutional Neural Networks, augments the understanding of a scene's 3D structure, capturing depth for precise distance estimation. Preliminary results from ongoing efforts, including the inception of a neural network for bounding boxes and aircraft classification, indicate the potential for improved aircraft tracking in real-world airport settings, offering fresh insights into computer vision-enhanced aviation tracking.
Did this research project receive funding support from the Office of Undergraduate Research.
No
Advanced Aerial Mapping with Computer Vision: Enhancing Aircraft Recognition, Location Estimation, and 3D Scene Comprehension using Monocular Depth Estimation and Deep Learning
In the realm of aviation, accurate aircraft recognition and localization are of paramount importance. This study delves into the application of computer vision in conjunction with aerial maps to enhance aircraft recognition, location estimation, and real-time tracking. The methodology focuses on developing an autonomous system for identifying and localizing aircraft, along with evaluating integrated computer vision against standalone methods. A significant part of the exploration focuses on monocular depth estimate (MDE), noted for its efficacy at extended ranges, cost efficiency, and simplicity. Merging MDE with deep learning, especially Convolutional Neural Networks, augments the understanding of a scene's 3D structure, capturing depth for precise distance estimation. Preliminary results from ongoing efforts, including the inception of a neural network for bounding boxes and aircraft classification, indicate the potential for improved aircraft tracking in real-world airport settings, offering fresh insights into computer vision-enhanced aviation tracking.