Detection of disease in fresh fruits using convolution neural network by comparing with support vector machine to maximize the accuracy and sensitivity

This study evaluates two machine learning methods for improved fruit disease diagnosis: one that uses innovative convolutional neural networks and the other that prioritizes accuracy and sensitivity improvement. From the many fruit disease photographs on Kaggle, three thousand samples were chosen. For the sake of testing, two datasets will be utilized: one containing 800 photos (or 30% of the total) and another including 1800 shots (or 70% of the total). Honesty, precision, and reliability by doing numerical calculations, we may assess how well the convolution neural network approach works. The novel CNN produced 97.5% accuracy, 5.1% sensitivity, and 99.4% precision, as opposed to the SVM method’s 83.3% accuracy, 87.9% sensitivity, and 88.3% precision. For both groups, the p-value is 0.000. It seems that the Novel CNN approach outperforms the SVM method when it comes to detecting fruit quality.