Parminder Singh

https://doi.org/10.64823/ijter.2606017

Starfruit (Averrhoa carambola) is a commercially important tropical fruit that is highly susceptible to various diseases, including anthracnose, fruit borer infestation, and bed bug damage, which significantly reduce yield and quality. Early and accurate detection of these diseases is essential for effective crop management and sustainable agricultural production. This study presents a deep learning-based approach using a custom Convolutional Neural Network (CNN) model for automated classification of starfruit diseases from image data. The proposed model is trained on a dataset comprising multiple classes, including Carambola Anthracnose Disease, Carambola Bed Bugs Disease, Carambola Fruit Borer Disease, Healthy Fruits, and Healthy Leaves. The CNN architecture is designed to efficiently extract spatial features and perform high-precision classification. Extensive experimentation shows that the model achieves exceptional performance with an accuracy of 99.80% and a near-zero loss, demonstrating highly stable learning and excellent generalization capability. The results indicate perfect or near-perfect classification across all categories, highlighting the robustness of the proposed model. This work confirms that custom CNN-based systems can significantly enhance automated plant disease detection and provide an effective solution for precision agriculture, enabling early intervention and improved crop health management.