This release contains the implementation of the CNN–XGBoost framework presented in the manuscript "A Hybrid CNN–XGBoost Framework for Real-Time Foreign Object Debris Detection in Aviation Safety Applications." Contents include: • ResNet-50 feature extraction implementation and training pipeline. • XGBoost classification framework. • Data preprocessing and augmentation scripts. • Training, validation, and testing procedures. • Performance evaluation and benchmarking scripts. • Explainable AI modules, including Grad-CAM and SHAP analysis. • NVIDIA Jetson TX2 deployment configuration. • Environment specifications and dependency files. • Reproducibility documentation and usage instructions. This version corresponds to the implementation used to generate the experimental results reported in the manuscript and is provided to support transparency and reproducibility of the research.