This study presents a methodology for constructing ensemble models лto solve recognition and decision automation tasks involving images from cargo and vehicle X-ray inspection systems. As a practical case, the detection of prohibited items is examined. The object detection networks are based on the YOLOv8x architecture, while EfficientNet-B6 is used for image classification. The performance of the models is evaluated using precision, recall, and average precision (AP) metrics, both for the individual detection networks and for the ensemble. The results highlight the advantages and limitations of the proposed approach. The methodology can also be extended to other recognition and decision automation tasks in cargo and vehicle X-ray inspection, such as identifying specific product categories, items, or their groupings.