부산대학교 도서관

Usually crops will get unused because of some plant diseases or with respect to transport issues and some storage facility. Greater than 15% of crops are left unused because of these diseases so it became a major issue to solve it. As many crops are getting wasted farmers need an automation system so that it could be helpful for the farmers to detect the plant diseases and help famers to overcome from it in prior to advance. Earlier Farmers used to detect the plant diseases by their eyes but it is not possible for all farmers to have a same way of recognizing the disease. With respect to the advance Artificial Intelligence concepts, we could able to adopt computer vision technique in the field of agriculture. We have a advance libraries available in the field of deep learning along with the user with developer creates a friendly environment to work together, with all these facilities make this concept to adopt in the field of agriculture to resolve the problem. As Image Classification is the base to this project, we have used a Deep learning concepts as it offers a easy working with the images. The concepts used in this project is taking the defect leaves and to title them with a particular disease pattern. With this defect images we will apply a pixel manipulation in order to get a useful information from the defected leaves. After this next step we have a feature extraction which can be done by image segmentation and finally we will do the classification of defected leaves depending upon the patterns that is taken from diseased leaves. For the purpose of classification of diseases, we will use CNN (Convolutional Neural Network) and for the demo purpose we have chosen public dataset which contains 87k images (RGB type images) which includes both healthy as well as diseased leaves.