Recent Advances of Computer Vision-based Plant Disease Recognition Methods

In the global agroecosystem, plant diseases, as important factors affecting crop yield, quality and ecological balance, have long been the focus of agricultural scientific research and practice. With climate change, the increase in international trade and the transformation of agricultural production methods, the frequency and distribution of plant diseases and the extent of their damage have shown a trend of increasing complexity and intensification. This not only poses a serious challenge to food security, but also poses a potential threat to sustainable agricultural development and biodiversity conservation. The aim of this paper is to provide an overview of current research advances and methods that can be applied to computer- based recognition of plant diseases and pests, as well as an outlook on future developments. The paper focuses on the recognition of plant diseases, categorizes the problem scenarios for recognition (single and multi-class) and in this way summarizes some informative methods (detection and tracking) and learning approaches for recognition.