Smart Robotic Pythomedic and Pesticide Sprayer using Image Processing and machine learning

Abstract

Plant-based diseases are a major concern all over the world and in India. Plant-based diseases lead to a reduction of the overall yield of the crop which leads to decreasing the overall income of the farmer. Every year, plant diseases cost the global economy a whopping $220 billion and estimate that 20-40 percent of the crops are lost annually from global food production. (Food and Agriculture Organization of the United Nations. (2019).  In the 21st century there are many technical and smart solutions available and can be made to help farmers in a better way, like traditional methods, smart disease detection using machine learning, farming robots etc but most of them are either expensive or difficult for farmers to implement and use. Our Solution is to design a very handy, easy to use solution for the farmers that can help them in identifying the problem that is there in the plant along with helping him smartly in solving the issue. The Smart Spraying robot has a camera in the front which captures the image of the plant and then applies the machine learning algorithm to detect the disease by analyzing the quality of the leaf. We have created our own data set on multiple vegetable plants. We have used a total of more than 3000 images to create our data set which includes three vegetables and in total 10 different leaf conditions.

The robot and the programs are tested in the lab for performance and accuracy. Our Machine learning model works with 100 percent accuracy in detecting the specific disease in the plant leaf. but as it is only trained on 10 different conditions its results are very limited as it can be used only on 3 different vegetable plants. It can be easily scaled to a very big level by making a new machine learning model which will include many more plants and their conditions. It is a viable and implementable solution in the farms which can provide great results.