Visit the GitHub Repo Fork.

Working Abstract⌗

Blending will be a major source of systematic uncertainty in downstream science analyses of LSST data. We aim to benchmark the performance of several deblenders, leveraging the Blending ToolKit (BTK) (Mendoza et al. 2024) to perform rigorous, end-to-end testing. This effort will involve key deblending algorithms, including SExtractor, Scarlet, BLISS, and DeepDISC, with the goal of comparing their effectiveness in handling blended galaxy images from LSST/Rubin simulations. BTK provides a robust simulation framework based on GalSim, capable of generating realistic galaxy blends under varying observing conditions. By utilizing BTK’s ability to create customized, reproducible blends, we will systematically test these deblenders against different blending conditions, such as source separation and brightness. The toolkit’s standardized evaluation metrics, including detection precision, segmentation accuracy, and shape reconstruction, will allow for comprehensive assessments of each algorithm’s strengths and limitations. Additionally, this project will incorporate challenging cases such as unrecognized blends, where blended galaxies are detected as single objects, pushing the deblenders to their limits. The results of this benchmarking will provide valuable insights into the performance of existing deblenders and highlight areas for future development.