Potato Leaf Disease Classification Using Deep Learning Model

India’s primary industry is agriculture, which suffers an annual loss of 35% agricultural yield due to plant diseases. Illness-related harvest losses are a serious issue for both major farming operations and rural communities. Subsequently, the detection of plant diseases is crucial to agriculture. If adequate care is not taken in this area, it could have a significant negative influence on plants by lowering the productivity, quality, and quantity of the corresponding good or service. Automatic disease detection not only reduces labor costs associated with maintaining vast fields of crops, but also picks up symptoms as soon as they appear on plant leaves. The majority of plant illnesses may be identified from the symptoms that occur on the leaves; however, due to the wide variety of diseases, recognizing and classifying diseases with the naked eye is not only laborious and time-consuming, but also prone to inaccuracy with a high error rate. In this study, authors proposed a sequential deep learning model where in each convolution layer is followed by a max pool layer in order to extract most relevant features form the input images. For experimental validation of proposed deep learning model, study uses 2152 potato leaves images from Plant Village Dataset out of which 1000 are of early blight and 1000 are of late blight the remaining 152 images are of healthy leaves. Authors have divided this dataset into 32 different batches and trained the model using multiple subsequent 2-Dimensional convolutional layers and 2-Dimensional Max pooling layer with Rectified Linear Unit (RELU) as the activation function. With ADAM optimizer and 50 epochs, authors achieved a maximum accuracy of 98.83% and a loss of only 4.47%.