Date of Award


Degree Type


Degree Name


Degree Program

Electrical Engineering


Electrical Engineering

Major Professor

Dr. Abdul Rahman Alsamman


Surface water is a critical resource that requires constant monitoring due to drastic impacts of human use, climate change, and severe weather phenomena. Existing monitoring techniques are limited and vary in their efficacy. It is proposed, here, that contour detection in the visible range is better for monitoring dynamic and long-term changes to surface water bodies. For that purpose, a semi-automated method for collecting and labeling water contours from Landsat-8 and Sentinel-2 images is presented. Due to the need for human inspection, the method has thus far generated 14K labeled images from more than 200,000 images. Given the cost of data labeling, a deep semi-supervised self-learning system is proposed, which performs learning in two stages, known as the teacher-student model. The teacher is trained on the accurate human-labeled data, then used to pseudo-label the remaining unlabeled data. The student is trained on both human-labeled and machine pseudo-labeled data. A uniquely designed multi-scale UNet classifier that uses fewer parameters and is developed and shown to be more accurate than other state-of-the-art (SOTA) classifiers for both teacher and student. Random augmentations are used to ”noise” the student model and improve its generalization, and normalization schemes are used to blend the human-labeled loss with the machine-labeled loss. Without the self-training, the multi-scale UNet classifier has 69.2% F-score over the SOTA systems, that improves to 73.58% with self-training.


The University of New Orleans and its agents retain the non-exclusive license to archive and make accessible this dissertation or thesis in whole or in part in all forms of media, now or hereafter known. The author retains all other ownership rights to the copyright of the thesis or dissertation.