Comparative Analysis Of Plant Disease Detection Models: Toward Sustainable And Environmental-Friendly Agriculture

Abstract

The timely plant disease detection is important for the sustainable environment and yield protection. This paper presents a comparative analysis of the hybrid method with the advanced deep learning (DL)methods for early detection of diseases in plants. Accurate and early diagnosis plays a crucial role in minimizing crop losses and enhancing agricultural productivity. By reducing the overuse of pesticides and enabling targeted interventions, these models contribute to more sustainable farming practices. The study highlights how intelligent disease detection can support environmental conservation by lowering chemical runoff and reducing resource waste. This study focuses on evaluating the performance of five advanced machine and DL models such as Vision Transformer (ViT), ResNet-50, EfficientNet-B5, and Xception alongside a proposed hybrid approach that combines ExtraTrees with K-Nearest Neighbors (KNN) for plant disease classification. The objective is to assess each model’s classification accuracy, computational efficiency, and resource consumption for selecting the most appropriate model in agricultural applications. The proposed ExtraTrees + KNN hybrid approach shows the highest overall efficiency and delivering strong classification accuracy while maintaining minimal computational overhead. Vision Transformer (ViT) achieves competitive results in terms of accuracy but requires comparatively more resources. ResNet-50 and EfficientNet-B5 also perform well. Xception performs moderately in predictive performance. The hybrid model’s capability makes it suitable for deployment in resource-constrained agricultural environments where computational resources are not enough as required by other deep learning models. This research is also highlighting the importance of evaluating both predictive performance and resource efficiency for sustainable plant disease monitoring and management system.

Abstract from the paper’s OpenAlex record; open copy licensed cc-by. Read at the source ↗

The text the model read, with the source record’s inline markup (italics, sub- and superscripts) rendered and its character entities decoded; quote offsets refer to the raw text.