iSAID 航空图像实例分割数据集
收藏资源简介:
iSAID 是一个用于航空图像中实例分割的大规模数据集。它是第一个用于航空图像中实例分割的基准数据集,结合了实例级物体检测和像素级分割任务。数据集包含 15 个类别中 655,451 个对象实例的 2,806 张高分辨率图像。 iSAID 的图像与 DOTA-v1.0 数据集相同,数据源自 Google Earth 、 JL-1 卫星及 GF-2 卫星(中国资源卫星数据和应用中心的)拍摄。
iSAID is a large-scale dataset for instance segmentation in aerial imagery. It is the first benchmark dataset for instance segmentation in aerial scenarios, combining instance-level object detection and pixel-level segmentation tasks. The dataset contains 2,806 high-resolution images with 655,451 object instances across 15 categories. The images used in iSAID are identical to those from the DOTA-v1.0 dataset, and the data is sourced from Google Earth, the JL-1 satellite, and the GF-2 satellite (provided by the China Resources Satellite Data and Application Center).




