article · IEEE Access
Effective agricultural management relies heavily on early identification of plant leaf diseases to protect crop yields and food supply chains. An automated ensemble deep learning framework categorises illnesses affecting apple and maize plants by merging traditional texture analysis with deep neural features. The architecture employs a hybrid preprocessing method, combining Local Binary Pattern texture descriptors with features drawn from two pre-trained convolutional neural networks. To enhance model efficiency, the resulting feature vectors are refined using three meta-heuristic algorithms: the Binary Dragonfly algorithm, Ant Colony Optimization, and the Moth Flame Optimization algorithm. Final classification is performed using high-performing machine learning classifiers. The integrated pipeline achieves an accuracy of 99.8 per cent, surpassing prevailing state-of-the-art techniques and establishing an effective computational strategy for early crop disease detection.
Plant diseases reduce harvest yields and fruit quality, posing direct risks to food security and farm incomes. Rapid automated identification allows agricultural managers to intervene promptly and curtail crop loss. Delivering an accuracy rate of 99.8 per cent, the reported methodology shows how blending computer vision techniques can make automated disease diagnosis significantly more reliable for major staple and fruit crops.
This methodology could underpin diagnostic software for farm management platforms, agritech scouting drones, or mobile applications used by crop consultants and farmers. The primary targets are apple and maize producers seeking automated crop health tracking. Currently, the technology appears to be applied and tested in an experimental computer vision setting, remaining several steps away from real-world field deployment where variable outdoor conditions must be managed.
AI-generated from the published abstract. Always read the original work before citing.
A stable and sustainable food supply chain is only possible with effective agriculture management. The application of technology in agriculture has recently produced encouraging outcomes in terms of improving agricultural yield and quality. Early detection and characterization of plant leaf diseases, which have a substantial influence on agricultural output and the general well-being of farmers and consumers, is a crucial component. The automatic identification and categorization of illnesses affecting apple and maize plants is the focus of this paper’s investigation into the implementation of cutting-edge technical solutions, notably machine learning algorithms. Deep learning has recently made significant contributions to the automatic detection and classification of plants diseases specifically in fruits and vegetables. As a result, the production and quality of fruits and vegetables have both improved. Many diseases have a negative impact on crop quality and output. This research proposes an ensemble deep learning based technique for classifying plant leaf diseases. In this work, two separate pre-trained deep learning models have been utilized and a hybrid approach is proposed. In this research work we have proposed a hybrid framework based on the hybrid preprocessing algorithm, an ensemble features engineering phase based on the texture features and two types of deep features extraction. The extracted Convolutional Neural Network (CNN) features are selected and fused with the LBP features. The ensemble feature vector is optimized using three different meta-heuristic algorithms that are Binary Dragonfly algorithm (BDA), Ant Colony Optimization algorithm and Moth Flame Optimization algorithm (MFO). The optimized feature vector is classified using state-of-the-art machine learning algorithms. Notably, the study addresses the research gap by providing a comprehensive analysis of plant diseases affecting fruits and vegetables. Additionally, it acknowledges the limitations of the proposed approach, paving the way for future research and improvements. The utilization of the proposed technique yields a remarkable accuracy of 99.8%, surpassing current state-of-the-art methods. This research contributes to the ongoing efforts to enhance food security and sustainability by providing an effective solution for the early identification and classification of plant diseases, thereby mitigating the negative impact on crop productivity and quality.
This page summarises published work. The authoritative version sits with the publisher.
DOI: 10.1109/access.2024.3389648
Is something wrong with this record? Report it or request removal.
Discussion
Have you built on this work, tried to replicate it, or seen it applied in practice? Share what you know. Verified researchers and MARATTO™ domain experts can open a discussion, and any member can reply. Contributions are reviewed before they appear.
No discussion yet. Open the first thread.
New to MARATTO™? Create a free account.
