Dataset and Software Tools for Shipwreck Detection from Sonar Imagery – UROP Spring Symposium 2024

Dataset and Software Tools for Shipwreck Detection from Sonar Imagery

Sabrina Lin

Pronouns: She/her

Research Mentor(s): Katherine Skinner
Research Mentor School/College/Department: Robotics / Engineering
Program:
Authors: Katie Skinner, Sabrina Lin
Session: Session 5: 2:40 pm – 3:30 pm
Poster: 43

Abstract

Countless pieces of history are scattered across the waters in the form of shipwrecks. They are important containers of cultural exchanges and technological advancements throughout the centuries. These wrecks often remain untouched because they are difficult to detect and may pose hazards to marine navigators. The Field Robotics Group hopes to create a free plugin on an existing GIS software that can detect shipwrecks from images of sea beds. Using the bathymetric data collected in Thunder Bay and data hosted on the NOAA website, we generated images of underwater terrains and shipwrecks. These images will be fed to a machine learning model to train an AI to detect shipwrecks from the images. When the AI is ready, we will incorporate it into QGIS, an open-source GIS software, in the form of a plugin to allow for widespread and easy use. Our research would provide another option from the few machine-learning models for shipwreck detection currently available. Additionally, the model can also be a basis for detecting other objects in the waters such as sinkholes and trenches. The QGIS plugin can also be used for other applications in the marine archaeology and science fields. This would help other marine scientists and archeologists navigate the waters safely and efficiently.

Engineering

lsa logoum logo