Voxel51/AFO-Aerial_Floating_Objects
收藏资源简介:
AFO数据集是第一个用于训练机器学习和深度学习模型进行海上搜索和救援应用的开源数据集。它包含了40,000个手工注释的人和漂浮在水中的物体,这些物体在水中很多都是小尺寸的,难以检测。数据集由50个视频片段中的图像组成,这些图像是由各种无人机搭载的相机拍摄的,分辨率为1280x720到3840x2160。从这些视频片段中提取并注释了3647个图像,包含了39991个物体,并且被分为训练集、测试集和验证集。数据集以Darknet YOLO格式准备。
The AFO dataset is the first open-source dataset for training machine learning and deep learning models for maritime Search and Rescue applications. It contains 40,000 hand-annotated persons and objects floating in the water, many of which are small in size and difficult to detect. The dataset consists of images from 50 video clips captured by various drone-mounted cameras with resolutions ranging from 1280x720 to 3840x2160. From these video clips, 3647 images containing 39991 objects have been extracted and manually annotated, and are split into training, test, and validation sets. The dataset is prepared in Darknet YOLO format.




