ABSTRACT

Plants play a vital role in the survival of all organisms on Earth. Due to this fact, it is veryimportant to ensure that measures are taken to detect and mitigate any diseases on plants. Plant diseases are a major factor for crop losses in agriculture. This paper presents a Deep Learning approach to detect and classify plant diseases by examining the leaf of a given plant. In this paper, the classification is performed in multiple stages to eliminate possibilities at every stage, hence providing better accuracy during predictions. A YOLOv3 object detector is used to extract a leaf from the input image. The extracted leaf is analyzed through a series of ResNet18 models. These ResNet18 models were trained using transfer learning. One layer identifies the type of leaf and the following layer checks for the possible diseases that could occur in the plant.