A Novel Metaheuristic Algorithm Optimized Deep CNN-LSTM Model for Amla Disease Identification

The quality evaluation of agricultural goods are vital components in improving their vendibility and managing wastes. Fruit diseases present a persistent risk to producers and consumers. Precision agriculture can assist in making decisions on the management of fruit diseases by using computer vision-based approaches. Manual disease detection is highly challenging and it takes an excessive amount of effort, knowledge, and processing time. Deep learning is therefore used to better accurately diagnose diseases. In this work, it is aimed to present deep learning framework approach for quickly identifying the apparent defects in amla fruit. In the proposed algorithm, deep convolutional neural network model with marine predator optimization algorithm is used for optimizing the accuracy of disease detection in amla fruit. The structure of the deep convolutional neural network comprises components for feature extraction and optimum classification. The hyperparameters for training convolutional neural network layers are tuned by marine predator optimization algorithm. The efficiency of proposed model is evaluated and compared to other optimization strategies. The optimized deep learning model outperforms by providing greater classification accuracy when compared to the current approaches, according to numerical analysis, demonstrating its potential for accurate disease detection and classification.