Li, Vladislav (2024) Scene analysis for smart devices and immersive technologies. (PhD thesis), Kingston University, .
Abstract
Augmented Reality (AR) has emerged as an innovative technology with promising applications across various domains, including gaming, education, healthcare, and manufacturing. Enhancing the visual fidelity and efficiency of AR systems is crucial for delivering immersive and seamless user experiences. The AR systems rely on scene analysis techniques to obtain information about the surrounding real environment and use that information to superimpose digital 2D and 3D information. This thesis presents a novel scene analysis methodologies that integrates cutting-edge techniques for AR such as super-resolution, oriented bounding boxes, 3D bounding boxes, and data augmentation for FSL with energy-efficiency in mind. The thesis describes techniques of object detection model evaluation using modern tools and standards. Considering object detection, in the first part of the thesis, superresolution techniques are analysed as a solution to the distant object detection, leveraging deep learning architectures to enhance the resolution and detail of captured scenes.
Actions (Repository Editors)
 |
Item Control Page |