Improved Spider Monkey Optimization with Deep Learning Model for Tomato Leaf Disease Recognition

Tomato plant leaf disease recognition is a crucial feature of smart farming, as it allows early analysis and intervention to avoid the spread of diseases that devastate tomato crops. With the advent of machine learning (ML), automated disease recognition methods have become increasingly popular. This method involves gathering a database of images of either healthy or diseased tomato leaves. Image processing systems are then utilized to enhance and preprocess these images. Deep learning (ML) models, mostly convolutional neural networks (CNNs), are trained on this database to classify leaves as diseased or healthy depending on their visual characteristics. These methods offer quick and consistent assessments of leaf health. Therefore, this study proposes an enhanced spider monkey optimization with a DL model for tomato leaf disease recognition (ISMODL-TLDR) technique. The ISMODL-TLDR technique incorporates the DL models with a hyperparameter tuning model for tomato leaf disease recognition. Wiener filtering (WF) is initially used as a preprocessing stage to improve the leaf image quality and reduce noise. Besides, an EfficientNet model captures intricate patterns from the preprocessed images. To fine-tune the model's hyperparameters effectively, an improved spider monkey optimization (ISMO) model can be introduced, which intelligently tunes the hyperparameters. At last, the classification stage employs long short-term memory (LSTM), enabling the method to comprehend temporal reliabilities in leaf disease progression. The simulation analysis portrays the enhanced achievement of the ISMODL-TLDR technique in terms of classification accuracy. The ISMODL-TLDR technique holds great promise for sustainable agriculture practices, helping farmers make informed decisions and mitigate disease-related crop losses.