Overview of the development of AI dataset annotation

Abstract

With the continuous development of artificial intelligence, various deep learning algorithms need a lot of training of annotated data, and how to improve the efficiency of data annotation has become a research hotspot. This paper analyzes the development history of AI data set annotation, summarizes the general framework of AI data set annotation, summarizes three semi-automatic or automatic AI data set annotation methods, and compares and analyzes the advantages and disadvantages of the three methods.