LISA Traffic Sign

LISA交通标志数据集是一组包含美国交通标志的视频和带注释的框架。它分为两个阶段发布，一个阶段仅包含图片，一个阶段同时包含图片和视频。这些图像现在可用，而完整的数据集正在进行中，并将很快提供。
47 US sign types7855注释在6610帧上。Sign尺寸从6x6到167x168像素。从不同相机获得的图像。图像大小从640x480到1024x522像素不等。有些图像为彩色，有些图像为灰度。数据集的完整版本包括所有带注释的符号的视频。每个符号都带有符号类型，位置，大小，遮挡 (是/否)，在路边 (是/否)。所有注释均以纯文本保存。csv-files.Includes一组Python工具来处理注释并轻松地从数据集中提取相关符号。

The LISA Traffic Sign Dataset is a collection of videos and annotated frames containing U.S. traffic signs. It was released in two phases: one phase only includes images, while the other contains both images and videos. The image-only portion is currently available, whereas the full dataset is still under development and will be released soon.

The dataset features 47 types of U.S. traffic signs, with a total of 7,855 annotations across 6,610 frames. The size of individual signs ranges from 6×6 to 167×168 pixels. All images are captured from various cameras, with their resolutions varying from 640×480 to 1024×522 pixels. Some images are in color, while others are grayscale.

The complete version of the dataset includes videos of all annotated signs. Each annotated sign is accompanied by metadata including sign type, bounding box position, dimensions, occlusion status (yes/no), and roadside presence status (yes/no). All annotations are stored as plain-text CSV files. The dataset also provides a set of Python tools for processing annotations and easily extracting relevant traffic signs from the dataset.