Transfer Learning-based Areca Nut (Areca catechu) Disease Detection using CNN and SVM Approaches with ResNet-50 for Improved Deep Learning Performance

Background: Fruit rot among Arecanut (Areca catechu) cultivation presents severe risks to agricultural yield together with product quality and agricultural profit margins. The timely detection of diseases along with accurate detection needs to be performed for minimizing economic losses and enabling sustainable agriculture systems. Methods: The study implements deep learning through transfer learning of the pre-trained ResNet-50 model. A method flows from ICAR Hirehalli proprietary data processing through Python software coding to Keras and TensorFlow implementation for model training. In addition to the assessment the study evaluated CNN along with SVM for traditional benchmarking purposes. Result: The efficacy of ResNet-50 during training reached 98% accuracy while validation accuracy settled at 92.76% surpassing both CNN with 90.83% accuracy and SVM with 89.95% accuracy. After training the model reached a substantial loss level of 0.2 which proved learning efficiency as validation loss settled at 0.3 indicating robust generalization.