Real Time Tomato Plant Leaf Disease Detection Using Convolutional Neural Network

Abstract

A substantial threat in achieving a good harvest is crop diseases. To achieve better results, the field of leaf-based image classification requires the advancement of precise approaches. The dataset is created with 200 pre-processed images. The model training includes creation of dataset, further clearing of noise in the images and assigning their corresponding pixel values. After uploading the test image, passed through Convolution, pooling, flattening and full connection are steps in the feature extraction process. The pre-trained model is run after loading an image, scaling it, adjusting the value of the pixels. The convolution neural network (CNN) used for classification of real-time leaf disease of tomato plants. The solution employs a convolution neural network with 97.5 ^% accuracy to identify disease in real time tomato plant leaves with in a span of 4 seconds.