Rotten Fruits and Vegetables Detection: A Deep Learning and Transfer Learning Based Approach

Identifying spoiled food is crucial due to the severe health risks associated with consuming it, such as food poisoning, which can cause symptoms like nausea, vomiting, and diarrhea. This condition can lead to dehydration and, in severe cases, require medical intervention and hospitalization. Vulnerable groups such as children, elderly, pregnant women, and immunocompromised individuals are particularly susceptible. Pathogens in spoiled food can also lead to long-term gastrointestinal issues. Ensuring food freshness not only promotes good health but also prevents discomfort and potential dangers from foodborne illnesses. In our study, various convolutional neural networks (CNNs) and transfer learning models-including MobileNet, Inception, Xception, DenseNet121, ResNet, EfficientNet, RegNetx, and NASNetLarge-were evaluated for their ability to distinguish between rotten and fresh fruits and vegetables. ResNet achieved the highest accuracy of 0.9930, while CNN models reached a peak accuracy of 0.9835, demonstrating their effectiveness in food quality assessment.