Computer vision and machine learning applied in the mushroom industry: A critical review

Abstract

BackgroundMushrooms are popular food items containing numerous vitamins, dietary fibers, and a large number of proteins. As a result, mushrooms can increase the body’s immunity and prevent many types of cancer to keep the body healthy. For these reasons, the demand for high yields and safety in the production of high-quality mushrooms is increasing. Scope and approachThis review highlights the application of computer vision and machine learning algorithms in the mushroom industry. Through a systematic review of papers published between 1991 and 2021, this article introduces key aspects related to mushrooms (e.g., species identification and quality classification based on artificial intelligence), and discusses the advantages and disadvantages of various approaches. Key findings and conclusionsNumerous artificial intelligence and machine vision technologies have been implemented in research efforts focusing on edible fungi. However, their applications are generally limited to the identification of poisonous mushrooms according to their forms, the plucking of cultivated mushrooms covered by soil, and the mechanized grading of mushrooms. Clearly, the currently available methods cannot meet the requirements of the digitization and intelligentization in the field of edible mushrooms. Considering these reasons, it is possible to develop further application opportunities, such as digital mushroom phenotype determination, and high-throughput breeding based on big data, and mechanical picking by a harvesting robot as well. Therefore, the integration of computer vision and machine learning with the development of more efficient algorithms will undoubtedly be a hotspot for future studies in the context of the mushroom industry.