A Deep Net Approach for the Segmentation & Detection of Infected Regions in Plant Leaves

Agriculture is the main source of income of about more than half of the population of India. Farmers work hard to produce different crops but the plant diseases contributes in the reduction of the production as the farmers are not aware of the diseases occurred to different plants. Different techniques have been designed till now to detect the diseases at the early stage so that the production can be enhanced by taking preventive measures. A recent study analyzed the infected mango leaves by the disease (Anthracnose) that is one of the fungal diseases. Although the existing model is capable of providing better results but there are some limitation such as a large number of processing layers was utilized that increased the complexity. Thus, to cope with the issues, a novel model using histogram equalization and region of interest is proposed. The novelty of using Histogram equalization (HE) technique increased the efficacy of the model. The simulation is performed in the MATLAB and the results successfully surpassed the existing system in terms of accuracy, missing report rate and false report rate.