Abstract

This study presents a hybrid MobileNet–Inception convolutional neural network for efficient tomato leaf disease classification in precision agriculture. The model integrates depthwise separable convolutions from MobileNet with multi-scale feature extraction using Inception modules to achieve a balance between accuracy and computational efficiency. A dataset of 5,000 tomato leaf images across 10 classes was compiled from Mendeley Data and Kaggle and split into training, validation, and testing sets in an 80:10:10 ratio. The model was trained for 50 epochs using the Adam optimizer (learning rate = 0.001), with dropout and L2 regularization applied to improve generalization. Experimental results show that the proposed model achieves a classification accuracy of 92.6%, with macro-precision, recall, and F1-score of 0.94, 0.93, and 0.93, respectively, outperforming benchmark models including AlexNet, VGG16, VGG19, InceptionV3, DenseNet121, ResNet50, MobileNet, and EfficientNetB0. In addition, the model demonstrates low inference latency (154 ms; 6.49 FPS), making it suitable for real-time applications. Analysis of confusion matrices indicates minor misclassifications among visually similar disease classes. These results highlight the model’s effectiveness and practical applicability for real-time tomato disease detection, supporting timely decision-making in precision agriculture systems.