Smart Detection and Diagnosis of Plant Disease Using Deep and Machine Learning Methods

Abstract

In a time of lower crop yields and rampant plant diseases, techniques like deep learning and machine learning aided by developed computer vision help increase the crop yield and classify, prevent as well as cure plant-related diseases. In the literature review section, we have quoted many instances of the CNN model and methods such as transfer learning, apple forest and others for the detection and classification of diseases on different crops with their spectacular results in numeric form. We have relied on quantitative secondary data to bring new datasets with unique results. Besides, we jotted down a brief introduction to some Machine and deep learning models. In the next and most important section, we evaluated the accuracy of these pre-trained learning models and classical machine learning algorithms by adding thousands of images of plants infected with different plant diseases, dividing these images into training and Test sets by 70/30 proportion, and by using metrics of Performance, precision, accuracy, and recall. Before conclusion, the accuracy and loss were determined by applying every image from the test dataset to each iteration.