SPS Seminar - Classifying planetary surfaces using machine learning

When:  Thursday 11^th February at 14.00

Where:  Microsoft Teams - Online

Speaker:  Alex Barrett - OU

Hosted by: Jack Wright

Abstract:

We are living in a golden age of robotic space exploration, where we have more high-resolution images of planetary surfaces than ever before. This gives us the ability to observe surface processes and landforms in unprecedented detail. In the case of heavily instrumented bodies such as the moon and Mars, the ever-increasing coverage of LROC-NAC and HiRISE data provides the potential for truly global studies. However, studies combining high resolution and large spatial extents present significant challenges.  Mapping or surveying large areas at full resolution becomes massively time consuming or requires very large teams to do effectively.

One solution is using machine learning systems to classify images by semantic segmentation. This approach can provide a useful tool to augment the workflow of human geomorphologists, “triaging” extremely large, high resolution datasets to identify concentrations of certain textures or landforms. I will discuss the process of developing classification systems for lunar and martian case studies, and how this approach can be applied to different science goals, each with their own unique challenges.

Bio:

“I am a planetary geomorphologist with a background in interdisciplinary environmental science. I did my undergraduate and master’s at Lancaster, studying environmental science with a focus on landscape geomorphology, planetary environments, and volcanology. I then did my Ph.D. at the OU and have been around in some capacity ever since. My Ph.D. research involved identifying periglacial landscapes on Mars, by means of remote sensing, field, and laboratory work. I have been compiling training datasets for AI studies since 2018, initially with a focus on the ExoMars landing sites. My current project is focused on mapping impact melt flows around lunar craters.”