This chapter presents an explainable framework for multi-angle license plate detection in autonomous traffic systems, leveraging an enhanced YOLOv8 architecture with oriented bounding boxes (OBBs). To ensure high accuracy and real-time performance, the model integrates a lightweight MobileNetV3 backbone and a shuffle attention mechanism, enabling more effective feature extraction and robust detection of small, rotated, and partially occluded plates. The proposed architecture achieves inherent interpretability by leveraging internal attention maps, eliminating the need for external gradient-based methods and offering intuitive insight into the model’s focus. Experimental results demonstrate that our model outperforms the standard YOLOv8 baseline, achieving a mAP@0.5 of 93.2% (a 7.3% improvement), with real-time inference at 33 FPS. These results highlight the model’s suitability for deployment in intelligent transportation and digital twin scenarios where explainability, efficiency, and reliability are essential.