Repository for the CRUMB dataset of FR galaxies
CRUMB (Collected Radiogalaxies Using MiraBest) is a machine learning image dataset of Fanaroff-Riley galaxies creating by merging MiraBest, FR-DEEP, AT17 and the supplementary MiraBest Hybrid dataset. All sources are cross-matched to ensure no duplicates images are present.
CRUMB consists of a total of 2100 images split into a training set of 1800 and a test set of 300. Images have two labels: a "basic" label - 0 (FRI), 1 (FRII) or 2 (hybid) - loaded as the dataset's default target, and a "complete" label which allows the user to select the class label from any of four the parent datasets. Please see the notebook for information on how to access these.