Abstract: Maize farming is a staple food source and economic backbone for farmers in West Africa. However, recent years have seen a decline in maize yields due to diseases. To address this challenge, lightweight models for early disease detection in maize plants were developed, intended to assist farmers through a user-friendly application. The dataset provided by the University of Cape Coast, Ghana, consists of six classes: five diseases and one healthy class. The study explored the effectiveness of ResNet9 and EfficientNet4 models for disease classification. Results analysis reveals that ResNet9 achieved an accuracy of 98.44%, while EfficientNet4 achieved an accuracy of 94.3%. Both models exhibit stable accuracy and minimal overfitting as validated through cross validation. These models were subsequently used to develop an application, integrated into the application's backend, with the model showing the highest confidence being used to display the detected disease. The application allows users to upload multiple images, making it applicable to real world scenarios. Through this application farmers can get immediate diagnostic feedback and remedy suggestions. Additionally, it offers an information hub to educate on the symptoms and causes of the diseases. This tool highlights the potential to advance agricultural technology in West Africa.