An Optimized Deep Learning Model EfficientNetB3 for Potato-Tomato Disease Detection
Year: 2026
Type: conference-paper
Abstract: Plant diseases are a major impediment to the farming yield, especially in crops of economic importance like potato and tomato. This paper suggests a streamlined deep learning model using EfficientNetB3 to classify leaf diseases with multiple classes. The model employs ImageNet-pretrained weights, which are used with customization of classification layers, data augmentation and Adamax optimizer to enhance generalization and convergence. On a PotatoTomato dataset based on the PlantVillage repository of six disease classes and healthy classes, experiments are carried out. The model proposed has the overall classification accuracy of 99.25, and high performance in all categories, including difficult to diagnose late-stage diseases. These findings indicate that EfficientNetB3 is an efficient and effective algorithm in automated crop disease detection in precision agriculture.
Language: English
Topic: Smart Agriculture and AI
Subfield: Plant Science
Domain: Life Sciences
SDG: Zero hunger
Open Access status: hybrid
OpenAlex ID: W7204787979