Fusing Deep CNN and Local Binary Pattern for Accurate Plant Disease Diagnosis

In today's world where the demand for food in-creases exponentially, a major thrift has occurred in the agri-culture industry to plant those seeds which are convenient and hassle-free to grow. Plants which have the possibility of being infected and which are difficult to cultivate are generally left behind. This in turn will lead to surplus amount of a certain commodity whereas produce scarcity for the others. Moreover, it will affect the economy as a whole. One of the major reasons for this shift is the lack of knowledge and expertise to tackle the unfamiliar diseases that occur in their cultivation. Farmers that are planting new crops with no prior experience will need proper expertise to identify diseases and their treatments for a better yield. But in the present scenario that will have severe overhead costs and thus culminates a lot of risk for the cultivator. In this scenario, an easy to use software that can analyse the issue and give a likeable solution is much desired. Hence, we propose our idea on Leaf Disease Classification and Remedies which can analyse the disease present and suggest probable remedies as well. Our system is built on the traditionally available CNN Models trained on the Field Plant data set to give a better accuracy of the in-field images. During identification, if the possibility of disease is above a threshold then a warning for the same is given. A locally-procured database consisting of treatments for various diseases were made to provide information to users about the identified diseases. And we have also incorporated a predictive analysis system which can (to some extend) give an idea about any possible chances of infection.