Optical Remote Sensing Image Understanding With Weak Supervision: Concepts, methods, and perspectives

In recent years, supervised learning has been widely used in various tasks of optical remote sensing image (RSI) understanding, including RSI classification, pixel-wise segmentation, change detection, and object detection. The methods based on need a large amount high-quality training data, their performance highly depends the quality labels. However, practical applications, it is often expensive time consuming to obtain large-scale data sets with labels, which leads lack sufficient information. some cases, only coarse-grained labels can be obtained, resulting exact supervision. addition, information obtained manually may wrong, accurate Therefore, understanding faces problems incomplete, inexact, inaccurate information, will affect breadth depth applications.