The article presents a deep learning-based approach for detecting early blight and late blight in potato leaves using the EfficientNetB0 model. Potatoes are a crucial crop in India, but leaf diseases like early blight and late blight significantly impact yield. Traditional methods for disease detection are limited, and farmers often lack the tools to identify these diseases early. This study addresses this issue by employing the EfficientNetB0 model, which uses width, depth, and resolution scaling to enhance classification accuracy. The dataset consists of 2152 potato leaf images, with 2000 images of diseased leaves and 152 of healthy leaves. The model achieved a testing accuracy of 99.05%, outperforming many existing techniques. The study also compares the performance of the EfficientNetB0 model with the ResNet50 V2 model, showing that the EfficientNetB0 model provides higher accuracy in detecting early blight, late blight, and healthy leaves. The model's performance is evaluated using metrics such as accuracy, precision, recall, and F1-score. The results demonstrate that the EfficientNetB0 model is highly effective for potato leaf disease detection, offering a promising solution for farmers to identify and manage plant diseases efficiently. The study highlights the potential of deep learning in improving agricultural productivity and disease management practices. Future work includes developing a user-friendly web application for disease detection and management.The article presents a deep learning-based approach for detecting early blight and late blight in potato leaves using the EfficientNetB0 model. Potatoes are a crucial crop in India, but leaf diseases like early blight and late blight significantly impact yield. Traditional methods for disease detection are limited, and farmers often lack the tools to identify these diseases early. This study addresses this issue by employing the EfficientNetB0 model, which uses width, depth, and resolution scaling to enhance classification accuracy. The dataset consists of 2152 potato leaf images, with 2000 images of diseased leaves and 152 of healthy leaves. The model achieved a testing accuracy of 99.05%, outperforming many existing techniques. The study also compares the performance of the EfficientNetB0 model with the ResNet50 V2 model, showing that the EfficientNetB0 model provides higher accuracy in detecting early blight, late blight, and healthy leaves. The model's performance is evaluated using metrics such as accuracy, precision, recall, and F1-score. The results demonstrate that the EfficientNetB0 model is highly effective for potato leaf disease detection, offering a promising solution for farmers to identify and manage plant diseases efficiently. The study highlights the potential of deep learning in improving agricultural productivity and disease management practices. Future work includes developing a user-friendly web application for disease detection and management.