DIOR
收藏资源简介:
DIOR数据集是一个大规模、公开可用的光学遥感图像目标检测基准,包含23463张图像和192472个实例,涵盖20个目标类别。该数据集在目标类别、实例数量和总图像数量上都是大规模的,具有广泛的目标大小变化,不仅在空间分辨率上,还在对象的内部和外部类别大小变异性上。此外,图像在不同的成像条件、天气、季节和图像质量下获得,具有高度的内部类别相似性和内部类别多样性。
The DIOR dataset is a large-scale, publicly available benchmark for optical remote sensing image object detection. It contains 23,463 images and 192,472 instances, covering 20 distinct object categories. This dataset is large-scale across three core dimensions: the number of object categories, the total number of instances, and the total count of images. It exhibits extensive variations in target sizes, which stem not only from differences in spatial resolution across the images, but also from intra-class and inter-class size variabilities of the objects. Furthermore, the images are acquired under diverse imaging conditions, weather scenarios, seasons and image quality levels, and feature high intra-class similarity as well as rich intra-class diversity.

- 1Object Detection in Optical Remote Sensing Images: A Survey and A New Benchmark · 2019年



