Özet:

Abstract crop diseases pose a serious death trap to food safety, but their rapid disease diagnosis remains burdensome in many parts of the world due to the lack of the necessary foundation. These days, deep learning models have shown better performance than hi-tech machine learning techniques in several fields, with computer vision being one of the most noticeable cases. Agronomy is one of the domains in which deep learning concepts have been used for disease identification on different parts of plants. Having a disease is very normal and common, but prompt disease recognition and early avoidance of crop diseases are crucial for refining production. Although the standard convolutional neural network models identify the disease very accurately but require a higher computation cost and a large number of parameters. This requires a model to be developed which should be efficient and need to generate less no of parameters. This research work proposed a model to identify the diseases of plant leaves with greater accuracy and efficiency compared to the existing approaches. The standard models like AlexNet, VGG, and GoogleNet along with the proposed model were trained on the Night shed plants leaf which is available in the plant village. It has 9 categorical classes of diseases and healthy plant leaves. A range of parameters, including batch size, dropout, learning rate, and activation function were used to evaluate the models' performance or achievement. The proposed model achieved a disease classification accuracy rate of 93% to 95%. According to the findings of the accuracy tests, the suggested model is promising and may have a significant influence on the speed and accuracy with which disease-infected leaves are identified.

Anahtar Kelimeler: