The UFPR-ALPR dataset contains 4,500 fully annotated images (containing more than 30,000 license plate characters) of 150 cars from real scenes. These images were acquired with three different cameras (GoPro Hero4 Silver, Huawei P9 Lite, iPhone 7 Plus) and are provided in PNG format with a size of 1,920 × 1,080 pixels.

40% of this dataset is used for training, 40% for testing, and 20% for validation. Each image is annotated with the following text: the camera that took the image, the location of the vehicle, other information such as vehicle type (car or motorcycle), manufacturer, model and year; the identification and location of the license plate, and the location of its characters.