Gibson Scene dataset

该数据集名为Gibson场景数据集，旨在为现实室内环境中的点目标导航任务提供训练和评估资源。该数据集包含了一个训练集，样本数量在50,000到400,000之间，以及一个包含10,000个样本的验证集，这些数据均由一个理想智能体生成。其规模覆盖了多个场景（72个训练场景和14个验证场景），专门用于点目标导航任务。

This dataset, named Gibson Scene Dataset, is designed to provide training and evaluation resources for point-goal navigation tasks in realistic indoor environments. It includes a training set with the number of samples ranging from 50,000 to 400,000, and a validation set containing 10,000 samples, all of which are generated by an ideal AI Agent. It covers multiple scenarios, specifically 72 training scenarios and 14 validation scenarios, and is tailored for point-goal navigation tasks.