In India, agriculture is vital due to the country’s expanding population and increasing food demands. Crop diseases pose a significant risk to food security, and early detection is crucial for improving production. This study explores machine learning and deep learning methods for detecting potato leaf diseases. Logistic Regression, K-Nearest Neighbors, and Random Forest achieved accuracies of 82.83%, 90.23%, and 84.91%, respectively, but struggled to capture complex patterns in image data. Their respective precision, recall, and F1 scores were as follows: LR (82.20, 82.83, 82.44%), KNN (90.03, 90.23, 89.93%), and Random Forest (79.49, 84.91, 82.09%). The Convolutional Neural Network (CNN) outperformed all models with an accuracy of 98.86%, a precision of 82.12%, recall of 82.36%, and an F1 score of 82.21%. The CNN’s ability to analyze images directly from pixel data demonstrates its potential for real-time disease diagnosis, helping farmers optimize resources, reduce manual inspection, and prevent crop losses. This study highlights the transformative potential of deep learning in enhancing plant disease detection, contributing to sustainable agriculture and food security.