Research Data and Code for "A Mission-Oriented Explainable AI Metric Framework for Autonomous Military UAV Decision Support"

This repository contains the research data and supporting code associated with the article “A Mission-Oriented Explainable AI Metric Framework for Autonomous Military UAV Decision Support.” The deposited materials support two components of the study. First, they document the two-stream systematic evidence review used to derive the proposed mission-aware XAI evaluation framework. Stream A contains screening and evidence records for general XAI evaluation research, while Stream B contains screening, full-text eligibility, and operational-requirement evidence for military, UAV, aviation, autonomy, cybersecurity, and related mission-critical contexts. Second, the repository contains the data and code supporting the UAV image-based recognition worked example. These materials include dataset split manifests, model training history, held-out classification results, perturbation-level explanation-robustness results, latency measurements, and Python scripts used for ResNet-18 training, final classification evaluation, Grad-CAM generation, controlled perturbations, and robustness-metric calculation. The VisDrone2019-DET images are not redistributed in this repository. Users should obtain the original dataset from the VisDrone project. The deposited manifests identify the samples used in the worked example and support reconstruction of the experimental subsets from the original dataset. The repository is intended to support transparency, reproducibility, and verification of the results reported in the accompanying article.