This is the dataset used in the paper AutoSourceID-Light (paper) for Training, Test and Validation sets. AutoSourceID-Light (ASID-L) is an innovative framework that uses computer vision techniques that can naturally deal with large amounts of data and rapidly localize sources in optical images. The dataset is made of 3 fields images taken with the MeerLICHT telescope: _20200601_191800_ (Omega Centauri globular cluster), _20210401_173445_(Fornax galaxy cluster) and _20210910_022724_ ('empty' CDF-S). For each field, 2 files are provided: the optical image (_red_cosmics_nobkgsub.fits) and the catalog of EDR3 Gaia sources that match the location of the fields (_GaiaEDR3_cat_SNR_redimage.fits). If you use the dataset, please cite the appropriate DOIs. Code available at https://github.com/FiorenSt/AutoSourceID-Light (Zenodo repository)