Technological support for detection and prediction of plant diseases: A systematic mapping study

Abstract

The field of plant disease diagnosis and epidemiology seeks to assess symptoms caused by pathogens. Different infectious and non-infectious agents can cause similar symptoms in plant organs. Diagnosing diseases is crucial, but it remains an inherently manual and error-prone task. Many works have been proposed to diagnose plant diseases, mainly using machine learning approaches. Even though this field affects agribusiness areas, little has been done to classify and map the current literature. This article presents a comprehensive overview of the current literature, and draw some research gaps, trends, and challenges that are worth investigating. A systematic mapping of the literature was carried out in pairs, following well-established practice guidelines. In total, 56 primary studies were carefully selected from a sample of 668 papers, which were retrieved from 9 widely recognized electronic databases. They were analyzed and categorized to answer seven research questions. The results show that 41% of primary studies applied machine learning techniques to detect diseases, 32% used image sensors to identify symptoms related to plant diseases, 30% focused on proposing new models of machine learning to detect diseases 34% were evaluation studies, and 71% were published in scientific journals. The association between computer vision and neural networks appears as a promising field of research for the detection of diseases. Finally, this article can serve as a starting point for upcoming studies, providing insights from a systematic map of the literature.