<![CDATA[Tomatoes are one of the most widely cultivated and consumed crops globally, but they are highly susceptibleto various diseases that can significantly reduce yield and quality. Early detection of thesediseases is crucial for effective management and prevention. The objective of this study is to developan accurate early detection system for tomato diseases using deep learning to support effective cropmanagement. The research method employed is a modified Convolutional Neural Network trainedon the PlantVillage dataset, which consists of 21,000 images across 10 disease classes. The studyevaluates three training scenarios using different epoch values (25, 50, and 75) to optimize modelperformance. Data preprocessing included image resizing and augmentation, followed by ConvolutionalNeural Network training and validation. The study’s results showed that increasing epochsimproved the model’s accuracy: 98.18% at 25 epochs, 98.53% at 50 epochs, and 99.19% at 75 epochs.Precision, recall, and F1-score also increased, from 90.95% at 25 epochs to 95.80% at 75 epochs, indicatingenhanced model reliability. However, longer training times were required as the epoch countincreased. This research concludes that a modified Convolutional Neural Network can accuratelyclassify tomato diseases, providing a reliable and practical tool for early disease detection. The proposedsystem has the potential to be integrated into mobile applications for real-time use in the field.It contributes to sustainable agriculture by enabling timely disease intervention and improving cropproductivity.]]>
Copyrights © 2025
