Abstract:?The investigation and remediation of the "illegal construction, extraction, cluttering and occupation (four illegal acts)" problems of the Yellow River is one of the indispensable links to ensure the high-quality development of the Yellow River Basin. In order to address this problem, this paper proposes a deep learning-based image target detection algorithm for the Yellow River "four illegal acts". Firstly, the image dataset of the target area is obtained from Google Earth; secondly, the sample library belonging to the "four illegal acts" problem is constructed; finally, the Unet algorithm is used to train the sample dataset and obtain the experimental results. The experimental results show that the accuracy of this method for the detection of the "four illegal acts" targets reaches 0.962, which can effectively improve the efficiency of monitoring and investigation, and provides a strong theoretical support for the "four illegal acts" remediation. This innovative method provides a new way to solve the problem of "four illegal acts" in the Yellow River Basin and is expected to play an important role in practice.
Keywords:?Key words: Four illegal acts;?Yellow River;?Unet;?Detection;?Deep learning
