Proposed Deep Learning Approach for Plant Diseases Detection

Farmers in most agricultural countries usually experience high annual losses due to crop diseases and plant infections, which threatens global food security and worsens the famine situation. Researchers have focused on trying to detect the earliest symptoms of tomato plant diseases due to the multilateral effort to find the potential cure to further prevent plant decay cases and decrease the food shortage risk worldwide. On the global market, tomatoes are one of the most grown crops and offer substantial value from sales. In this paper, we are proposing a model based on a convolutional neural network and using the Very Deeper VGG16 architecture for tomato plant disease detection and classification. We have employed our method to identified and classified ten classes, nine for various disease types and one for healthy tomato leaves, a dataset of 16,012 images of tomato leaves from PlantVillage with ten diverse categories has employed. The comparison results showed that VGG16 achieved the highest accuracy of 98%, outperforming the CNN by 2.2%. And we compared the results with some recent related works.