A comparative study of various techniques for crop Disease Detection and segmentation

Plant diseases have always been a major concern with food security in particular, the quality assurance of the agricultural products usually depends on the appropriate identification of the diseases, if any, in the crops used in their production. Moreover, since a vast majority of world's economy usually depends on the agriculture, it becomes more important to cut short on the losses incurred with the traditional methods of crop disease detection that involves human supervision and observations taken with naked eyes which are often inaccurate, time consuming and error prone. Since food production and its consumption could have consequence that spans multiple spheres ranging from a country's economic growth to health of its denizen, it becomes more than relevant to establish a rather reliable and automatic process for crop disease detection. This study proposes a comparative study underlying the various researches involving computer vision in particular, Machine Learning and Deep Learning, for crop disease detection automation, as a part of underlying framework, this study intends to present the necessary methods and techniques such as acquiring data, feature extraction, image segmentation, etc., that withstood the test of the time in the required context.