Answers are grounded in this paper and cite their sources.
The increasing demand for agricultural productivity is challenged by plant diseases that have led to crop losses and significant financial impacts. Early detection of plant diseases can save critical resources and prevent further crop damage, ensuring sustainable agricultural practices. However, existing methods for plant disease classification often face limitations, including high computational costs, resource constraints, and internet dependency, especially in rural agricultural regions. To address these challenges, we propose a novel edge–cloud collaboration framework that integrates the optimized CNN model for accurate and energy-efficient plant disease classification. The framework leverages edge devices for local real-time processing to reduce latency and energy consumption while utilizing the cloud for computationally intensive tasks to ensure scalability and robustness. The proposed framework provides a collaborative processing mechanism in which the edge-enabled servers execute the feature extraction and preliminary classification tasks based on raw input while allowing the cloud server to aggregate and refine the global model. The customized CNN is of a lightweight design that can flexibly integrate both feature extraction and classification layers to achieve high accuracy with relatively low resource consumption. Key contributions in this work are a scalable architecture uniquely adapted to resource-constrained settings, an efficient CNN architecture for plant disease classification, and extensive evaluations across diverse datasets. Simulation results demonstrate that the proposed framework achieves up to 99.3% accuracy with 0.998 F1-score for the PlantVillage dataset, 97.12% accuracy with 0.971 F1-score on the Corn or Maize dataset, and 99.45% accuracy with 0.994 F1-score on the Potato leaf disease dataset. These results confirm the effectiveness of the proposed framework and its contribution to sustainable agricultural practices by enabling real-time, resource-efficient plant disease detection across various environments.
No takes yet. Share an insight, caveat, or question.
Akram et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: