Turmeric leaf disease detection using optimized deep learning model

Turmeric is popular for its medicinal, culinary, and non-culinary uses. Turmeric cultivation is affected due to disease in their leaves. Farmers conventionally identify the leaf diseases and provide pesticides to avoid further loss. However, manual identification of leaf disease over large farms is time-consuming and requires human effort. Also, in several types of leaf disease, the specific type that occurs in the plant should be distinguished; thus, proper knowledge of farming is required. To simplify this and to know the type of disease, smart agricultural practices are followed worldwide. One of the intelligent techniques adopted in farming is leaf disease detection. By providing the filed image to a learning algorithm, the type of disease can be detected accurately without manual interventions. ML and DL are used in this smart agricultural practice. To improve the performance further, an optimized deep learning model is presented in this research work. A Novel gazelle optimization algorithm is combined with deep learning MobileNetV3 network to attain enhanced disease detection performance. Experimentation on our own dataset with 5000 samples, including leaf spot, leaf blight, leaf rot, and curl disease-affected leaves, the proposed model performance is evaluated and observed with a maximum accuracy of 96.8%, a precision of 97.1% over conventional disease detection models.