The Potato and Tomato is cultivated worldwide and ranks as the fourth largest food crop. Since fungi are the primary cause of potato crop infection, both early and late blight infections affect them. Plant diseases have a big impact on output, hence crop diseases need to be found and understood. Utilizing deep learning, smart farming enables the automated identification of sick crops. Disease management and control in real-time improve output and lower agricultural losses. In this paper, two distinct CNN (Convolutional Neural Network) architecture that is suitable for detecting potato and tomato illness is proposed. The training set's database is built via image processing. Categorical Crossentropy is employed for model analysis, while Adam is used as the optimizer. The last judgment function is Softmax. Accuracy is maintained while the convolution layer and resources are minimized. The experimental findings demonstrate the suggested model's 98.14% and 91.34% accuracy in detecting plant illness respectively.