Radio Source Catalogue Generated with YOLO-CHARS for SKA SDC1 at 560 MHz

Description This record contains the catalogues associated with the YOLO-CHARS framework for automated radio source detection and source characterisation on the SKA Science Data Challenge 1 (SDC1) simulated continuum images. The dataset includes the following catalogue files: S_M_L_merge_catalog.csv – Complete merged catalogue generated by the YOLO-CHARS framework, containing the predicted source properties produced by the detection and characterisation models. SSD_catalog_standalone.csv – Standalone source-detection catalogue generated by the SSD branch of the YOLO-CHARS framework. ALL_Merged_catalogue_v1.csv – Initial merged catalogue containing all predicted source parameters. best_score(93.57).csv – Catalogue corresponding to the highest overall SDC1 score obtained in this work. best_purity(99.53).csv – Catalogue corresponding to the highest source purity obtained in this work. purity_comparison_with_pybdsf_same_subfield(99.43).csv – Catalogue used for the source-purity comparison between YOLO-CHARS and PyBDSF on the same SDC1 subfield. PyBDSF_catalog.txt – PyBDSF catalogue used for the comparative analysis presented in Appendix C of the associated manuscript. The catalogues are provided to support the reproducibility of the published results and to facilitate further analyses. The S_M_L_merge_catalog.csv file contains the complete set of predicted quantities generated by the YOLO-CHARS framework, including source centre coordinates, core coordinates, integrated flux density, core fraction, source size, shape parameters, detection confidence, model-branch outputs, and bounding-box dimensions. Users wishing to evaluate the catalogue using the official SKA Science Data Challenge 1 scoring procedure should note that the Conf, M_Clas (or Clas), W, and H columns in S_M_L_merge_catalog.csv are not required by the official scoring pipeline and may be removed prior to scoring. If you use these catalogues in your research, please cite both this Zenodo dataset and the associated publication.