Evaluating Machine Learning Approaches for Detecting Coronal Mass Ejections in Images

At a Glance:

• Computer Vision (CV), a form of Machine Learning (ML), can successfully detect CME in solar images (SOHO/LASCO C2)

• COTS ML models, when trained with data processed using novel image processing techniques, perform better than prior bespoke models in the task of identifying CME.

• Use of COTS ML models provides a significant savings in needed resources (CPU time, Training dataset size) which makes many more potential problems solvable using CV.