Synthetic Aperture Sonar Images Generation using Neural Networks
Abstract
This Master’s thesis focused on the development and implementation of a neural network, specifically a conditional diffusion model for the generation of SAS images required to train neural networks for object classification and detection on the seabed. This topic arises because only a limited amount of data is available for this purpose, due to the high costs and complexity of the systems used to acquire such data. Therefore, it is necessary to increase and generalize the amount of available data in order to properly train neural networks for classification and detection tasks. This thesis first provides an overview of the state of the art, including acoustic systems, particularly sonar systems and synthetic aperture sonars, neural networks, and diffusion models. The datasets, model, and methodologies are then described. Finally, the results obtained with the conditional diffusion model are analyzed. The adopted approach consists of harmonizing only masked regions (i.e., the object region and its shadow) while keeping the real background unchanged. The results were evaluated through a test involving several users, who were asked to observe a set of images and determine whether they were real or AI-generated. The results indicate that fairly good performance was achieved, although there is still room for improvement both in terms of image quality and data quality, for example regarding the datasets used to train the model.
Read the paper
Related studies
- Learning from First-Generation Qualitative Approaches in the IS Discipline: An Evolutionary View and Some Implications for Authors and Evaluators (PART 1/2)
- Commentary on the clinical and preclinical dosage limits of interstitially administered magnetic fluids for therapeutic hyperthermia based on current practice and efficacy models
- Prosthesis: New Technological Opportunities and Innovative Biomedical Devices
- The history and future of neural modeling for cochlear implants
- Ryanodine receptors: waking up from refractoriness
- Novel Transgenic Rabbit Model Sheds Light on the Puzzling Role of Matrix Metalloproteinase-12 in Atherosclerosis
Go deeper in Evidence Hub
Evidence Hub adds a study data sheet (design, sample size, evidence level, key numbers), an AI summary and a chat grounded in the paper's text.