Fruit Type Classification using Stacked Bi-directional Long Short Term Memory

In recent years, Machine Learning (ML) and Deep Learning (DL) applications have become increasingly important in agriculture, particularly in the classification of fruits based on image data. Fruit recognition and categorization have attained a lot of attention from researchers. Because of the diversity and heterogeneity of fruits, this is a difficult challenge. To solve this, a Stacked Bi-LSTM image classification approach has been proposed to enhance the accuracy of fruit image classification. The process includes feature extraction by a Convolutional Neural Network (CNN), followed by feature selection using a Recurrent Neural Network (RNN). Following that, the proposed Stacked Bi-LSTM model is utilized to classify fruit images, resulting in enhanced accuracy. As indicated by evaluation parameters like as recall, precision, and F-measure, the acquired results outperform previous algorithms. The analyzed outcomes show that the Stacked Bi-LSTM model accomplishes precision of 85.4%, recall of 88.3%, and F1-score of 83.9%, demonstrating greater performance when compared to existing approaches.