Systematic Diagnosis of Cross-Domain Performance Degradation in PCB Defect Detection: A PKU-Source Study — Data, Analysis Outputs, and Reproducibility Resources

This repository provides the data, analysis outputs, protocols, source data, and reproducibility resources supporting the study “Systematic Diagnosis of Cross-Domain Performance Degradation in PCB Defect Detection: A PKU-Source Study.” The study uses PKU as the sole source domain and MIXED-clean, SELF, and DeepPCB as observational target domains under a common five-class ontology consisting of mouse_bite, open_circuit, short, spur, and spurious_copper.

The repository contains the study-specific dataset manifests and ontology mappings, experimental and evaluation protocols, annotation-audit and adjudication records, full-domain evaluation outputs, positive-control results, controlled-imaging and selected-target analyses, domain-characterization and robustness outputs, statistical results, and source data underlying the supplementary tables. It also provides the custom scripts, notebooks, configurations, environment specifications, and other resources required to trace and reproduce the reported analyses.