Classification of Corn Leaf Disease Detection Using ResNet50 and Inception V3

I Indarti (Universitas Nusa Mandiri)
Dewi Laraswati (Universitas Bina Sarana Informatika)
F Frieyadie (Universiti Kuala Lumpur Malaysia)

Journal Medical Informatics Technology
Volume 4 No. 2, June 2026

Corn is an important food crop in Indonesia, requiring accurate classification methods to support agricultural productivity. To evaluate and compare ResNet50 and Inception V3 models with augmentation techniques for corn image classification. Deep learning classification using CNN architectures with rotation, shifting, and flipping augmentation. 3,560 corn images (2,500 training, 700 validation, and 360 testing). ResNet50 achieved 93.05% accuracy, while Inception V3 achieved the highest performance with 94.02% accuracy, 93.04% precision, 93.00% recall, and 93.02% F1-score. Image augmentation significantly improved classification performance, and Inception V3 was identified as the most effective model for corn image classification.