병해충 피해 인식을 위한 딥러닝 활용

To recognize the plant disease, the Deep Learning is utilized. 12 layers of CNN architecture have created. RMSProp has used for optimization and used ReLU as Activation function. In case of Pear fire blight and pear scab, those diseases have similar symptoms. When datum are unbalanced, those similar symptoms occurred significant performance impact. To cover lack of data, we used data augmentation that could increase datum. As a result, those datum could improve the performance.