Abstract
The growth of the Indian economy is mainly based on agricultural production. A method for crop disease detection based on pre-processing and segmentation processes using filtering and neural network techniques is proposed. The dataset here has been collected based on the pre-historic cultivation data and disease-affected data of the crop. Live images from the field have been collected and the dataset has been created. This data has been initially processed using a pre-processing technique based on convoluted Gaussian filtering. Then the processed image has been segmented using a deep active contour convolutional neural network (DACCNN) to formulate new loss functions which incorporate the region and information about size in the disease detection while training. From the results of the experiment, the proposed method is a vigorous method for crop disease detection and also segments main diseases of plant leaves like Cercospora Leaf Spot, Bacterial Blight, Powdery Mildew, and Rust.