Agrobot

Abstract

Green pepper has a very vast market in Pakistan, it is cultivated in vast areas of South Punjab and Sindh. In 2007 Pakistan was among the top five chili pepper producers in the world after India China, Bangladesh, and Peru. (Agri Corner, 2015)But in last few years due to global warming, poor diagnosis or ignorance of either farmers or agriculture field personals the pepper production has gradually decreased especially in South Punjab where it has nearly become impossible due to recent white fly attacks to make this crop profitable. Field officers appointed by Agriculture department are either disqualified or outdated so there is a need to build a system that could help both farmers and agricultural officers to properly and accurately diagnose the plants so that crops can be saved from being destroyed. In this project we developed a system that would help farmers and agriculture experts to detect diseases and recommend relevant disease control mechanisms like pesticides to overcome the diseases by applying image processing techniques on datasets containing the images of healthy and diseased plants and artificial intelligence to recommend the disease control mechanisms. Image processing is used to detect whether the plant is healthy or diseased and after detecting that plant is diseased it classifies according to the type of disease like bacterial, fungal, viral etc. We have five classes in our model which is trained using the Tensor Flow on Google Collab, we have experimented many techniques for model training but ended up with CNN model as it gave us most accuracy we have achieved 78% training accuracy from some techniques but after using the CNN we got 95% accuracy while training. System also gives recommendations about the detected disease in plants like pest control mechanism recommendation another relevant information about plant health and its recovery. The data used to provide control mechanism is taken by the help of pepper plant researcher.