From MetaAI insight to the detection of rare echoes of stellar explosions

Rodiat Ayinde

Light echoes are the reflection of stellar explosions on interstellar dust. They offer valuable information in the study of dust and transients, but their detection is challenging as they appear as extended, morphologically diverse, time-evolving transients often at the limit of surveys’ signal-to-noise ratio. Our group is testing and developing AI models to automate the detection and study of light echoes in the Rubin LSST data. We tested a recent MetaAI model “Segment Anything”, or SAM (Kirillov et al. 2023), to assess its suitablity in the segmentation of template-subtracted astronomical images containing ligth echoes using the ATLAS dataset as a precursor of LSST.


This poster will be displayed on Monday and Tuesday.


Career Stage: 
Undergrad Student