Dynamic optimized distributed deep learning model for tomato leaf disease detection and classification

Tomato is the most popular vegetable, cultivated worldwide because they serve as a staple in global cuisines. Early classification and detection of tomato leaf disease substantially reduce yield loss and enhance the quality of fruit. Nevertheless, traditional techniques struggled with inaccurate detection results, lack of robustness, and computational complexity due to inconsistent architecture and input quality issues. Hence, the research developed a Dynamic Optimized Distributed Learning-based Convolutional Neural Network Long Short-term Memory (D2CNTM) for detecting tomato leaf diseases and classifying their types. Here, the model extracts the textural pattern and context-aware characteristics through analysing the inconsistencies using the symptoms. Moreover, the D2CNTM includes a distributed learning strategy, which performs parallel training to accelerate the training speed and reduce the computational overhead. Substantially, the utilization of the nature-inspired dynamic social foraging optimization (DSFO) algorithm finds the best solution to refine the parameters of the D2CNTM and offers superior convergence. Owing to these advancements, the D2CNTM attains maximum precision, specificity, sensitivity, and accuracy of 98.71%, 98.48%, 97.78%, 98.01% using the Tomato-Village Dataset.