Abstract: Variety purity is a crucial cereal seed quality indicator, and the safety and quality of seeds are directly tied to the security of the nation's food supply. The national standard for variety purity identification uses the phenol staining and morphological identification methods, and the results of the identification are time-consuming and subject to the inspectors' level of knowledge. Recently, cereal variety recognition, purity detection, and clarity detection have significant development, which is facilitated by machine vision technology, machine learning, and deep learning algorithms. The current research findings and issues in the areas of image acquisition, image pre-processing, machine learning, and deep learning technology in the field of cereal variety identification are summarized and analyzed in this review, and a forecast on the future research priorities in this area is also provided.