DV-PSO-Net: A novel deep mutual learning model with Heuristic search using Particle Swarm optimization for Mango leaf disease detection

Abstract

Efficient identification of diseases in mango leaves is crucial for maintaining the health and productivity of mango crops. Existing approaches, relying on manual inspection or image processing, are time-consuming, error-prone, and face challenges with various image conditions. This study addresses these issues by proposing a robust machine learning model capable of classifying mango leaf diseases across diverse conditions, including different resolutions, structural complexities, and varying blur levels. The solution involves exploring and optimizing machine learning algorithms, tuning hyperparameters, and developing a predictive model for accurate disease identification based on visual features extracted from leaf images. To overcome these challenges, a novel deep mutual learning model, DVNet, is introduced, leveraging the strengths of Densenet 121 and VGG19 neural networks. Hyperparameter optimization, a systematic procedure for identifying optimal values, is incorporated using Particle Swarm Optimization (PSO). The proposed framework achieves an impressive accuracy of 94.72% in detecting eight distinct disease categories and healthy mango leaves, surpassing existing works in mango leaf disease detection.

URN:NBN:sciencein.jist.2024.v12.804