Berkeley DeepDrive Video
收藏arXiv2025-09-30 收录
下载链接：
https://github.com/gy20073/bdd_driving_model/
下载链接
链接失效反馈官方服务：
资源简介：
该数据集是一个真实世界的数据集，用于在低数据环境下评估以对象为中心的模型。它包含了多样化的驾驶行为，用以评估以对象为中心的模型的性能。该数据集包含了约20万个测试帧，其任务是对自动驾驶策略进行评估。
This real-world dataset is designed for evaluating object-centric models in low-data regimes. It includes diverse driving behaviors to assess the performance of such models, contains approximately 200,000 test frames, and its core task is to evaluate autonomous driving policies.
提供机构：
Berkeley DeepDrive
