Efficient and accurate detection of tomato leaf diseases using fine-tuned EfficientNetV2 in Keras

Tomato is a widely consumed vegetable both in Senegal and worldwide. However, it is frequently affected by insect pests and diseases, leading to a significant reduction in yield. Curative and preventive approaches have been implemented for disease detection, but some are costly, while others remain prone to human error. In this study, we employ a Convolutional Neural Network (CNN) to classify tomato leaf diseases using deep learning techniques. The proposed approach involves preprocessing tomato leaf images, training an EfficientNetV2B3 model pre-trained on the ImageNet dataset for feature extraction, fine-tuning it on the tomato leaf dataset, and evaluating the model's performance on a test set. Experimental results show that the model achieves an overall accuracy of 94%, demonstrating excellent capability in identifying two specific diseases, Tomato Yellow Leaf Curl (TYLC) and early blight, as well as healthy leaf conditions. The proposed method offers a reliable and efficient solution for tomato disease detection, which is essential for ensuring food security and reducing financial losses in agriculture. The model performs well even in cases of severe infections, highlighting the potential of deep learning methods for automated and accurate disease classification in tomato cultivation.