adasdaadadad/Car_plate_OCR_dataset

Car_plate_OCR_dataset是一个包含约45,500张俄罗斯车牌图像及其文本标注的数据集，专门用于训练神经网络识别车牌号码。数据集基于Nomeroff Net项目，经过清理去除了不符合标注格式的图像。数据分为训练集（37,775张，83%）、验证集（4,891张，10.7%）和测试集（2,845张，6.3%）。标注信息直接体现在图像文件名中，包含拉丁字母和数字的组合，字符集为1234567890ABEKMHOPCTYX。

Car_plate_OCR_dataset is a dataset of approximately 45.5K images of Russian car plates and their text annotations, designed for training neural networks to recognize car plate numbers from images. The dataset is based on the Nomeroff Net project and has been cleaned by removing images that did not meet the annotation format. The data is divided into training (37,775 images, 83%), validation (4,891 images, 10.7%), and test sets (2,845 images, 6.3%). The annotations are provided in the image filenames, which consist of uppercase Latin letters and numbers, with the character set being 1234567890ABEKMHOPCTYX.