SAVH Plate Image Datasets for YOLOv8-CLS License Plate Legibility Classification and ANPR Field Validation

This record contains the plate image datasets used in the SAVH prototype, a vehicular counting and monitoring system designed for ANPR-based event capture, cloud processing, and visual license plate legibility classification.The record includes two complementary datasets. The first dataset is a relabeled version derived from LPLCv2 and organized for binary image classification using YOLOv8-CLS. The classes are placa_legible and placa_no_legible, representing visually legible and non-legible license plate evidence. This dataset was used for training, validation, and testing of the plate legibility classification model. The second dataset contains ANPR field plate images captured in Guayaquil using a Dahua ANPR camera. This dataset supports contextual and field validation of the SAVH prototype under real urban monitoring conditions, including variations in lighting, camera perspective, vehicle type, motion, and evidence quality. These datasets support the experimental validation of a cloud-based vehicular monitoring architecture that integrates ANPR event capture, visual evidence storage, YOLOv8-CLS inference, event traceability, and dashboard-based analysis. The datasets are provided to promote transparency, reproducibility, and traceability of the research results associated with the SAVH system. The data are intended for academic and research purposes related to license plate image quality assessment, ANPR systems, computer vision, intelligent transportation systems, vehicular monitoring, and smart city applications.