Grape Leaves Disease Identification using Transfer Learning Techniques

Because of the constantly shifting climatic and environmental circumstances, it is almost inevitable for crops to get infected with diseases. Crops are susceptible to a variety of diseases, which may hinder their development and production and are notoriously difficult to eradicate. It is vital to have accurate illness diagnosis as well as control activities that may be taken to avoid them in a timely manner in order to assure excellent quality and high productivity. Grape, which is a commonly cultivated crop in India, is susceptible to a broad variety of diseases. Leaf illness are the first signs that might appear as a result of fungal, bacterial, or viral infection. Therefore, it is necessary to create an automated system that can be utilized to determine the kind of sickness and then take the right measures. This paper has suggested an autonomous system that utilizes image processing and deep learning methodology for the purpose of diagnosing illnesses that affect grape plants. Using deep learning-based method, the characteristics were collected from the segmented sick area of the plant and then categorized as healthy, rotten, infected with ESCA, or affected by leaf blight. With the help of transfer learning method based on Resnet50, the accuracy of the test was improved to 99.58 percent.