부산대학교 도서관

This paper presents an agricultural aid application, developed and designed, to help farmers by utilizing Image Processing, Machine Learning and Deep Learning concepts. Our application provides features such as early detection of plant disease, implemented using various approaches. After evaluation, results showed that Convolutional Neural Network was performing better for plant disease detection with an accuracy of 97.94% at 20 epochs. It further helps the farmer to forecast the weather to decide the right time for agricultural activities like harvesting and plucking. To avoid reoccurrence of disease due to loss in soil minerals, a crop specific fertilizer calculator is incorporated which can calculate the amount of urea, diammonium phosphate and muriate of potash required for a given area. Since India is a multilingual country, the application has been designed to incorporate language translation in four languages: Marathi, Hindi, Punjabi and English.