基于深度学习的遥感影像地块分割方法

赵伟昱,张宏海,仲波

A Deep Learning Based Method for Remote Sensing Image Parcel Segmentation

ZHAO Weiyu,ZHANG Honghai,ZHONG Bo

表3 使用不同预训练数据集作为主干网络训练出的模型获得的预测图像的平均交并比

Table 3 Mean Intersection over Union of predicted images obtained from models trained using different pre-trained datasets as the backbone network

预训练数据集 平均交并比 MS-COCO（本文方法） 70.3% Imagenet 67.9% Cityscapes 68.5%