Guangzhou International Campus Autonomous Driving Dataset

该数据集包含从自动驾驶车辆视角捕获的图像，其特点是较低的摄像头角度，这与实际的自动驾驶应用，如送货和巡逻，紧密对齐。此外，该数据集被用于在各种驾驶场景下测试基准，并基于诸如注视、场景理解和逻辑等指标评估基于大型语言模型的技术。任务旨在利用大型语言模型评估自动驾驶系统在实际操作中的实时能力。This dataset comprises images captured from the perspective of autonomous vehicles, featuring a low camera angle that closely aligns with real-world autonomous driving applications such as delivery and patrol operations. Furthermore, this dataset is employed for benchmark testing across diverse driving scenarios, and for evaluating Large Language Model (LLM)-based technologies against metrics including gaze, scene understanding and logical reasoning. The task is designed to leverage large language models to assess the real-time operational capabilities of autonomous driving systems in practical scenarios.