Published Paper Details:

Classification of plant disease is important in reducing losses of yields, but the traditional diagnosis method is inaccessible to a large number of farmers. This paper assesses Plantix, which is a smart mobile application that employs image recognition using deep learning to detect diseases and pests in plants, nutrient deficiencies, etc. The backend is trained on huge annotated datasets, which allows it to classify 30+ crops and 400+ disorders using CNN-based models. Inference outputs are a disease classification, severity estimation, and the cause of the disease after image acquisition. The app also gives the treatment plans, chemical, biological, and cultural plans, as well as nutrient control, weather forecast, and crop calendar by season. Plantix has a diagnostic accuracy of over 90% but is affected by light, image sharpness, type of crop, and position of the symptoms. Although it has such merits as quick inference, multilinguality support, and sharing of the community, there are also obstacles, such as rare-disease coverage, the reliance on connection, and not being much integrated with local soil or sensor data. Altogether, Plantix has strong potential with regard to the applicability as an AI-powered and scalable crop-advisory system.