Allison Du
Pronouns: she/her
Research Mentor(s): Katherine Skinner
Co-Presenter:
Research Mentor School/College/Department: Naval Architecture and Marine Engineering/Robotics Institute / Engineering
Presentation Date: April 20
Presentation Type: External
Session: Session 2 – 11am – 11:50am
Room: Michigan
Authors:
Presenter: Table 2
Abstract
Marine robotic platforms have advanced in recent years to enable large-scale data collection of marine environments. Work in marine science including coral reef and fishery monitoring can advance with machine learning methods that leverage these large, labelled image sets. The goal of this project is to design and implement an interactive GUI to visualize and label data collected from marine robotic systems for use in machine learning applications on processing large datasets. The GUI enables visualization of existing labels, pixel-wise annotation masks, and other properties of the data for easy interpretation and analysis. Interactive features allow for manual annotation and generating segmentation masks from custom bounds. Data drawn from these annotations can be used in machine learning applications. Furthermore, the GUI’s cross-platform availability provides convenient access to users without a background in computer science. Overall, the GUI developed for this project will be an important tool for enabling machine learning research in marine science applications.
Engineering
