街景及交通标志图片数据集

该数据集由清华大学公开，包含约82097张带交通标识的腾讯街景全景图及16817张带标注的交通标志图像，总计近10万张图像。数据涵盖多种光照与天气条件，包含无标注训练图像，适用于半监督学习。主要用于自动驾驶中的交通标识识别任务，支持检测、分类与定位研究，提升模型在复杂环境下的鲁棒性。This dataset is publicly released by Tsinghua University. It consists of approximately 82,097 Tencent Street View panoramic images with traffic signs and 16,817 annotated traffic sign images, amounting to nearly 100,000 images in total. The dataset covers diverse lighting and weather conditions, and includes unlabeled training images, making it suitable for semi-supervised learning. It is primarily designed for traffic sign recognition tasks in autonomous driving scenarios, supporting research on detection, classification and localization, and enhancing the robustness of models in complex environments.

搜集汇总

背景与挑战

背景概述

该数据集是清华大学公开的街景及交通标志图片数据集，总计近10万张图像，包括带交通标识的街景全景图和带标注的交通标志图像，覆盖多种光照与天气条件，适用于半监督学习。主要用于自动驾驶中的交通标识识别任务，支持检测和分类，具有大规模、多样化的特点，适合研究和开发。