UT-DCR: Reproducibility Artifacts for Leak-Aware Evaluation and Uncertainty-Gated Reconciliation for Reliable Rice Disease Screening at the Edge

his record contains the reproducibility artifacts accompanying the study “UT-DCR: Leak-Aware Evaluation and Uncertainty-Gated Reconciliation for Reliable Rice Disease Screening at the Edge.” The repository includes the leak-free group-wise data split, evaluation fold indices, cached outputs from the frozen image classifier and lesion detector, soil measurements used in the evaluation, and analysis code required to reproduce the reported evaluation, calibration, and statistical analyses. The underlying rice-disease images originate from a third-party corpus and are therefore referenced by relative paths rather than redistributed.