부산대학교 도서관

Sericulture involves the production of mulberry silk from the stage of moriculture(cultivation of mulberry leaves) to silkworm reeling i.e., extraction of silk from the silkworm cocoon. Silkworms are prone to several diseases, which causes huge economic loss in silk production and early detection of infections in silkworms is an urgent issue that must be addressed right away. There is a definite need for automation in this field, especially during the larval stage. It has been proposed here to use CNNs to detect diseases. Early detection and identification of diseases would help farmers prevent diseases from spreading. Deep Learning has been used in this model to identify the disease in silkworms. For training the model various layers in the deep neural network is applied to categorize diseased and undiseased silkworm to obtain an accuracy rate. An accurate classification of infected and non-infected silkworms was achieved by CNN with a recall of 1.00 and 0.98, precision of 0.98 and 1, F1-scores of 0.99 and 0.99 for the diseased and undiseased classes, and an accuracy of 98.96%. The model was end-to-end trained on Tensorflow using Python language.