CODA
收藏资源简介:
CODA数据集是由华为诺亚方舟实验室创建的一个针对自动驾驶领域的真实世界道路角落案例数据集。该数据集精选了1500个真实驾驶场景,每个场景中平均包含四个对象级别的角落案例,覆盖了超过30个对象类别。CODA数据集的构建旨在解决自动驾驶系统在面对不常见对象和极端情况时的检测挑战,通过提供高质量的标注数据,推动自动驾驶技术向更可靠的方向发展。数据集的应用领域主要集中在自动驾驶系统的对象检测性能评估,特别是在处理复杂和不可预见的路况时的能力。
The CODA dataset is a real-world road corner case dataset for the autonomous driving domain, created by Huawei Noah's Ark Lab. It comprises 1500 carefully curated real driving scenarios, with an average of four object-level corner cases per scenario and coverage of over 30 object categories. Developed to address the detection challenges encountered by autonomous driving systems when facing uncommon objects and extreme scenarios, the CODA dataset aims to promote the advancement of more reliable autonomous driving technologies by providing high-quality annotated data. Its primary applications focus on the performance evaluation of object detection for autonomous driving systems, particularly their capability to handle complex and unforeseeable road conditions.




