DroneSURF

该数据集名为DroneSURF，包含了200个视频，涉及58个对象，在主动和被动监控设置下，通过无人机捕捉的41.1万个帧。其中，对象被分为40%用于测试，剩余60%用于训练和验证。该数据集规模宏大，200个视频带有超过786,000个人脸标注，其任务重点在于人脸识别。

The dataset named DroneSURF contains 200 videos covering 58 objects, with 411,000 frames captured by drones under both active and passive surveillance settings. Of these, 40% of the objects are allocated for testing, while the remaining 60% are used for training and validation. This dataset features a large scale, as the 200 videos are accompanied by over 786,000 facial annotations, and its key task focuses on face recognition.