Frontiers in plant science · 2026
Optimised hybrid late fusion deep learning model for cashew disease classification.
СтатьяНазвание и аннотация: Перевод пока не подготовлен
Обработка статьи · этапы, ошибки и управление
Загрузка состояния…
Ошибки и ограничения
История · было → стало
Состояние обновляется каждые 5 секунд, пока блок открыт. Процент показывается только там, где известен объём работы. Пауза останавливает продолжение этапа; текущий запрос к источнику может завершиться.
K Meenakshi, Sankaranarayanan Suresh
Аннотация
INTRODUCTION: Cashew farms are highly vulnerable to the attack of pests and plant diseases, which leads to massive losses in production and economic value. To diagnose the cashew disease, recent studies have explored deep learning methods; however, the existing models often have poor representative features and low generaliz-ability to different situations.
METHODS: As part of these solutions, we suggest a hybrid late-fusion architecture (combining EfficientNetV2-M and MobileNetV3-S) to extract the features and train them with XGBoost and CatBoost to give the classification, using the BOHB optimisation to make the hyper-parameter choices. By combin-ing MobileNetV3-S and EfficientNetV2-M, we can benefit from fine-grained visual information and computer efficiency.
RESULT AND DISCUSSION: The proposed model was experimented on the CCMT dataset comprising Anthracnose, Gummosis, Leaf Miner, Red Rust, and healthy leaf samples, achieving classification accuracies of 90% and 93%, with reduced computation times of 0.20 seconds for the XGBoost classifier and 0.07 seconds for the CatBoost classifier. As our findings show, boosting-based classifiers and efficient backbone networks can be combined to identify cashew disease effectively and computationally less complex.
METHODS: As part of these solutions, we suggest a hybrid late-fusion architecture (combining EfficientNetV2-M and MobileNetV3-S) to extract the features and train them with XGBoost and CatBoost to give the classification, using the BOHB optimisation to make the hyper-parameter choices. By combin-ing MobileNetV3-S and EfficientNetV2-M, we can benefit from fine-grained visual information and computer efficiency.
RESULT AND DISCUSSION: The proposed model was experimented on the CCMT dataset comprising Anthracnose, Gummosis, Leaf Miner, Red Rust, and healthy leaf samples, achieving classification accuracies of 90% and 93%, with reduced computation times of 0.20 seconds for the XGBoost classifier and 0.07 seconds for the CatBoost classifier. As our findings show, boosting-based classifiers and efficient backbone networks can be combined to identify cashew disease effectively and computationally less complex.