Deep Learning Based Image Overlap Detection
Trevor Landeen · Digital Commons - USU (Utah State University) · 2018
Deep learning is attracting a lot of attention because of its success in many research areas. This research is concerned with how deep neural networks may be used to process or compare multiple inputs with an abstract goal. Specifically, this paper addresses the case where the identification of an overlapping region between two images is desired. The scope of this paper includes the conceptual development of different approaches and a discussion of how observing a human’s approach to the same task can lead to improved performance. Some results are presented, but a thorough analysis and comparison of the trained DNNs are not included in this paper.