Disease classification in plants plays a major role in increasing the production of crops. Leaf diseases may seriously reduce yield and affect the quality of the pepper plant. Thus, the timely detection of diseases with a highly accurate method plays a key role in effective management and treatment and in reducing crop failure. Convolutional neural network (CNN) is considered the best tool for the automation of identification of plant diseases due to its automatic learning capability and feature extraction. This study investigates the effectiveness of transfer learning using MobileNet V2 and Inception V2 models for classifying bacterial spots in bell pepper plants. Bacterial spot disease significantly threatens bell pepper cultivation, necessitating accurate and efficient detection methods. Leveraging pretrained CNN architecture, MobileNet V2, and Inception V2, this research aims to identify the optimal model for disease classification in agricultural settings. Experimental results demonstrate that MobileNet V2 outperforms Inception V2 in classification with an accuracy of 99.96 percent.