Crop Disease Detection Systems: Techniques, Challenges and Trends

Early and accurate detection and diagnosis of diseases in crops is critical to ensuring global food supply, reducing agricultural losses and increasing agricultural productivity. Traditional methods, such as manual by visual professionals, are time-consuming, subjective and very easily made wrong. Due to the developments in recent technologies such as ML, DL, IoT and image processing, there has been a significant change in the crop disease detection domain. Considering that early and correct diagnosis of the diseases of a plant from multimodal data sources, including leaf images, environmental factors, and remote-sensing images, is crucial, such approaches are designed to assist to provide timely and accurate prediction. In this paper, give a complete review on the state-of-the-art in detecting crop disease using images, sensors and both of them. It also includes a commentary on the relative effectiveness of the approaches, examining their specific outcomes, constraints and suggestions. Meanwhile the paper also shares some lessons-learned and bottlenecks when deploying state-of-the-art algorithms in practice based on the practical application with limited datasets and generalizability of models. The paper concludes by proposing some potential ways forward to achieve scalable, generalizable, farmer led implementation of such solutions.