FoodScanAI: A dual-modal deep learning system for food spoilage detection using synthetic sensor data and image classification

Food spoilage is a worldwide problem affecting public health and the overall efficiency of food supply chains. Herein, a dual-model approach using environmental sensor data and visual inspection for food freshness detection is proposed. Synthetic sensor data were generated with a cGAN, while a fine-tuned VGG16 convolutional neural network performed image classification. The predictive performance of both models was high (F1-score ≥ 0.85), and the models were integrated into an easy-to-use web interface for real-time classification. This approach illustrates the feasibility of combining synthetic data augmentation and deep learning for practical, scalable food quality monitoring applications.