RodoSol-ALPR

该数据集称为RodoSol-ALPR数据集，包含由位于Rodovia do Sol (RodoSol) 特许经营公司拥有的pay tomers的静态摄像机捕获的20,000张图像，该公司在巴西圣埃斯皮里图州运营67.5公里的高速公路 (ES-060)。
有不同类型的车辆 (例如，汽车、摩托车、公共汽车和卡车) 的图像，在白天和晚上，从不同的车道，在晴朗和雨天，从车辆到相机的距离略有不同。所有图像的分辨率为1,280 × 720像素。
拟议数据集的一个重要特征是，它具有两种不同的车牌 (LP) 布局的图像: 巴西和南方共同市场 (为了保持与现有作品的一致性，我们将 “巴西” 称为巴西在采用南方共同市场标准之前在巴西使用的标准)。
每个图像在文本文件中都有以下信息: 车辆的类型 (汽车或摩托车)，LP的布局 (巴西或Mercosul)，其文本 (例如ABC-1234) 以及每个的位置 (x，y) 它的四个角。我们标记了拐角，而不仅仅是LP边界框，以便能够训练探索LP校正的方法，以及更广泛的数据增强技术的应用。
关于与我们的数据集有关的隐私问题，我们指出，在巴西，lp与各自的车辆有关，即，没有有关车辆驾驶员/车主的公开信息。此外，在每个图像中，所有人脸 (例如，驾驶员或RodoSol的员工) 都被手动编辑 (即，模糊)。

This dataset, named RodoSol-ALPR, contains 20,000 images captured by static cameras owned by the Rodovia do Sol (RodoSol) franchise company, which serves paying customers and operates a 67.5-kilometer highway (ES-060) in the Brazilian state of Espírito Santo.

The dataset includes images of various vehicle types (e.g., cars, motorcycles, buses, and trucks) captured under diverse conditions: daytime and nighttime, different driving lanes, sunny and rainy weather, and with varying distances between the vehicles and the cameras. All images have a resolution of 1,280 × 720 pixels.

A key feature of the proposed dataset is that it covers images with two distinct license plate (LP) layouts: Brazilian and Mercosul. To maintain consistency with existing research works, we define the "Brazilian" layout as the standard used in Brazil prior to the adoption of the Mercosul standards.

Each image is paired with a text file that contains the following information: the vehicle type (car or motorcycle), the LP layout (Brazilian or Mercosul), the LP text (e.g., "ABC-1234"), and the (x, y) coordinates of each of its four corners. We annotated the corner positions rather than just the LP bounding box to support the training of methods that explore LP rectification, as well as the application of more comprehensive data augmentation techniques.

Regarding privacy concerns associated with this dataset, we note that in Brazil, license plates are only linked to their corresponding vehicles, meaning no public information about the vehicle's driver or owner is available. Furthermore, all human faces in each image (e.g., drivers or RodoSol employees) have been manually anonymized via blurring.

搜集汇总

背景与挑战

背景概述

RodoSol-ALPR是一个包含20,000张巴西高速公路车辆图像的数据集，涵盖多种车辆类型和天气条件，特别标注了两种车牌布局及车牌位置信息，适用于车牌识别相关研究。