Plant disease identification and pesticides recommendation using Dense Net

Abstract

Plant diseases, mainly caused by bacteria and fungi, affect crop yield and quality. Detecting disease symptoms at an early stage and promptly is a significant obstacle in safeguarding crops. In developing nations, experts and agronomists commonly opt for visual identification of diseases on vast farms, which incurs both time and monetary expenses. Scientists have suggested diverse deep neural network architectures for recognizing plant ailments. Nevertheless, deep learning algorithms necessitate a vast amount of parameters, which extends the training duration but yields commendable precision. While deep learning and Densenet are widely used in pesticide recommendations. Researchers have suggested diverse deep-learning architectures for detecting agricultural ailments and recommend appropriate pesticides. Test images were diagnosed using an automated Densenet model and the results were verified by plant pathologists. An accuracy of over 92% was achieved in identifying the disease. Our solution is an innovative, scalable and accessible tool for disease management of various crops that can be implemented as a cloud service for farmers and professionals involved sustainable agricultural production.

Keywords

• Image processing
• genetic algorithm
• plant disease detection
• classification
• feature visualization
• crop diseases