This website requires JavaScript.

Semi-Supervised Domain Adaptation for Semantic Segmentation of Roads from Satellite Images

Ahmet Alp KindirogluMetehan Yal\c{c}{\i}nFurkan Burak Ba\u{g}c{\i}Mahiye Uluya\u{g}mur \"Ozt\"urk
Dec 2022
摘要
This paper presents the preliminary findings of a semi-supervisedsegmentation method for extracting roads from sattelite images. ArtificialNeural Networks and image segmentation methods are among the most successfulmethods for extracting road data from satellite images. However, these modelsrequire large amounts of training data from different regions to achieve highaccuracy rates. In cases where this data needs to be of more quantity orquality, it is a standard method to train deep neural networks by transferringknowledge from annotated data obtained from different sources. This studyproposes a method that performs path segmentation with semi-supervised learningmethods. A semi-supervised field adaptation method based on pseudo-labeling andMinimum Class Confusion method has been proposed, and it has been observed toincrease performance in targeted datasets.
展开全部
图表提取

暂无人提供速读十问回答

论文十问由沈向洋博士提出,鼓励大家带着这十个问题去阅读论文,用有用的信息构建认知模型。写出自己的十问回答,还有机会在当前页面展示哦。

Q1论文试图解决什么问题?
Q2这是否是一个新的问题?
Q3这篇文章要验证一个什么科学假设?