Label draws, scripts and per-image records for "With a foundation-model backbone, camouflaged object detection needs few labels except for small objects"

Label draws, training and evaluation scripts, per-image prediction records and result files for the paper "With a foundation-model backbone, camouflaged object detection needs few labels except for small objects" (C. Ma, P. H. N. Dang, Y. Wang; submitted to Image and Vision Computing). For each run, the archive also lists the prediction and decision rule written down before the run, with the outcome (preregistrations.csv). Model checkpoints and the COD10K, CAMO and NC4K datasets are not included; README.md describes the layout, the mapping from runs to the tables and figures of the paper, and how the scripts were run. Code is released under the MIT licence; the data files under CC BY 4.0.