IMPLEMENTASI METODE DEEP LEARNING DAN ALGORITMA CONVOLUTIONAL NEURAL NETWORK UNTUK KLASIFIKASI PENYAKIT PADA TANAMAN PANGAN MENGGUNAKAN RESIDUAL NETWORK

Food crops are plants that produce production and contain the main carbohydrates and proteins as a source of basic human food. Food crops can also be affected by several diseases, including pests and leafhoppers. This disease greatly disturbs the yields of these food crops which causes the income of farmers to decrease. The author makes a Machine Learning system to identify diseases in food crops, the aim is to help farmers in automatically identifying plant diseases so that their harvests can be more optimal. This system was created using the Deep Learning method and using the Convolutional Neural Network (CNN) algorithm and using the Residual Network (ResNet) model or architecture for the identification process of diseases in food crops. The way the CNN algorithm works is that it first breaks the image into several images, then inserts each smaller image into the small neural network, then stores the results of each small image into a new array, then downsamples, and makes predictions. The conclusion is that the CNN algorithm is able to produce a significant level of accuracy because it has a network depth and has been widely applied to image data. The result of this research is a machine learning system for the classification of diseases in food crops based on leaf images. Based on the results of the classification carried out, an accuracy rate of 94% was obtained with the best architecture using 90% dataset comparison scenario parameters, size 250x250 pixels, kernel 3x3, learning rate 0.010, optimizer Adam, epoch 2, batch size 32 and random seed 6.