LPLC
收藏资源简介:
LPLC数据集是一个用于车牌可读性分类的公开数据集,包含10,210张车辆图片,共有12,687个带有注释的车牌。这些图片覆盖了各种车辆类型、光照条件和相机/图像质量水平。数据集的注释策略包括车辆和车牌级别的遮挡、四种可读性类别(完美、良好、差和不可读),以及三个类别(不包括不可读车牌)的字符标签。LPLC数据集旨在支持车牌可读性分类研究,并可作为车牌检测和识别的基准数据集。该数据集对研究车牌识别技术中的图像质量与字符可读性之间的差异具有重要意义,并可用于评估现有车牌识别方法的性能。此外,LPLC数据集还包含夜间拍摄的照片,这在其他公开数据集中较为罕见,为研究夜间车牌识别提供了宝贵的数据资源。
The LPLC dataset is a public dataset for license plate readability classification, containing 10,210 vehicle images with a total of 12,687 annotated license plates. These images cover diverse vehicle types, lighting conditions, and camera/image quality levels. The dataset's annotation scheme includes occlusion annotations at both vehicle and license plate levels, four readability categories: Perfect, Good, Poor, and Unreadable, as well as character labels for three categories excluding unreadable license plates. The LPLC dataset is designed to support research on license plate readability classification, and can also serve as a benchmark dataset for license plate detection and recognition. This dataset holds significant importance for studying the discrepancy between image quality and character readability in license plate recognition technologies, and can be used to evaluate the performance of existing license plate recognition methods. Furthermore, the LPLC dataset includes photographs taken at night, which are relatively rare in other public datasets, providing valuable data resources for research on nighttime license plate recognition.




