Plant diseases pose a significant threat to global food security, impacting crop yield, and quality. To avert crop diseases from spreading, early detection is crucial. This study addresses the critical need for accurate and efficient crop leaf diseases classification by developing a custom Convolutional Neural Network (CNN) model. By utilizing a publicly available dataset of 87,867 images across 38 classes and 14 crop species, the proposed model achieved a training accuracy of 98.99%, a test accuracy of 98.38%, an overall ROC AUC score of 99.99%, and an overall average precision score of 99.88%. These metrics highlight the model’s robust performance and its potential utility in agricultural disease management. Furthermore, a detailed classification report indicates the model’s high precision, recall, and F1 scores across most classes, showcasing its effectiveness in capturing detailed features for accurate classification. This research provides a valuable foundation for developing advanced diagnostic tools aimed at promoting sustainable agricultural practices, thus contributing to improved crop health and productivity.