DrIFT
收藏资源简介:
DrIFT数据集是由渥太华大学和加拿大国家研究委员会共同开发的视觉无人机检测数据集,专门用于研究在域偏移条件下的无人机检测。该数据集包含14个不同的域,涵盖了视角、合成到真实数据、季节和天气等多种域偏移类型。DrIFT数据集通过背景分割图强调背景偏移,提供了背景相关的评估指标。数据集的创建过程结合了真实和合成数据,确保了数据集在训练和验证集中的平衡分布。DrIFT数据集主要应用于无人机检测和无人驾驶领域,旨在解决环境变化、视角多样性和背景变化对无人机检测性能的影响。
The DrIFT dataset is a visual drone detection dataset jointly developed by the University of Ottawa and the National Research Council Canada, specifically designed for research on drone detection under domain shift conditions. This dataset contains 14 distinct domains, covering multiple types of domain shifts including viewpoints, synthetic-to-real data, seasons and weather conditions. The DrIFT dataset emphasizes background shifts via background segmentation maps and provides background-related evaluation metrics. The dataset's creation process combines real and synthetic data to ensure a balanced distribution in both training and validation sets. The DrIFT dataset is mainly applied in the fields of drone detection and autonomous driving, aiming to address the impacts of environmental changes, diverse viewpoints and background variations on drone detection performance.




