DEEP NEURAL NETWORK BASED SMART OBSERVATION AND DISEASE DIAGNOSIS EXPERT SYSTEM FOR ONION CROP

Onion crop production is highly affected by diseases such as purple blotch, rust and smut, white tip disease, and downy mildew that affect foliage. The bulbs are often affected by the onion fly, white rot, stem and bulb eelworm, and neck rot. Agricultural information and deep learning are used for quick and accurate diagnosis of diseases in onion plants in this research. An automatic crop monitoring system with an imaging technique is presented where the onion field images are captured periodically for constructing the field monitoring system. The disease symptoms are detected by training the deep neural network model, and the performance of the proposed system is compared with the existing systems. The abnormalities observed in the plant are used for detecting the disease using the forward chaining approach in the proposed system in a real-time environment. The field monitoring system comprises an image logging module, a wireless transceiver, a motor system, and a camera. The deep learning model is trained with a limited amount of infected and healthy onion crops. The ambiguous boundary is used for the primary identification of crop disease symptoms. The proposed model provides a classification accuracy of 96.84% with faster speed when compared to the conventional techniques due to the fewer features used.