Related Experiment Videos
Optimised hybrid late fusion deep learning model for cashew disease classification
Meenakshi K1, Suresh Sankaranarayanan2
1Department of Networking and Communications, School of Computing, SRM Institute of Science and Technology, Kattankulathur, Tamil Nadu, India.
Frontiers in Plant Science
|June 8, 2026
Summary
This study introduces a novel hybrid deep learning model for cashew disease detection, achieving high accuracy with efficient computation. The model combines EfficientNetV2-M and MobileNetV3-S for feature extraction, offering a promising solution for agricultural pest and disease management.
Area of Science:
- Agricultural Science
- Computer Science
- Machine Learning
Background:
- Cashew farming faces significant economic losses due to pests and diseases.
- Existing deep learning models for cashew disease diagnosis exhibit limitations in feature representation and generalizability.
Purpose of the Study:
- To develop an effective and computationally efficient deep learning model for cashew disease identification.
- To address the limitations of current models in feature extraction and generalizability.
Main Methods:
- A hybrid late-fusion architecture combining EfficientNetV2-M and MobileNetV3-S for feature extraction.
- Training classifiers using XGBoost and CatBoost, optimized with BOHB for hyper-parameter selection.
- Utilizing the CCMT dataset, including Anthracnose, Gummosis, Leaf Miner, Red Rust, and healthy samples.
Main Results:
- Achieved classification accuracies of 90% and 93% for cashew disease detection.
- Demonstrated reduced computation times of 0.20 seconds (XGBoost) and 0.07 seconds (CatBoost).
- The hybrid model effectively identifies cashew diseases with high accuracy and computational efficiency.
Conclusions:
- Combining efficient backbone networks with boosting-based classifiers offers an effective strategy for cashew disease identification.
- The proposed model provides a computationally less complex solution for diagnosing cashew plant diseases.
- This approach enhances the potential for automated disease management in agriculture.