Monitoring the post-cooking quality of pasta fresca, especially filled types like ravioli, is challenging due to subtle, progressive changes during storage, driven by moisture migration and surface drying. In this work, we introduce a deep computer vision approach to discriminate quality variations in ravioli samples subjected to different time-temperature storage conditions. Leveraging a Residual Neural Network (ResNet) with transfer learning, we classify pasta samples into three degradation classes, simulating realistic storage scenarios. Our model achieves high accuracy in capturing visual cues correlated with water redistribution and texture changes. To ensure model transparency and facilitate adoption in food technology pipelines, we incorporate EXplainable Artificial Intelligence (XAI) techniques, which highlight key image regions driving classification decisions. The results demonstrate the feasibility of using deep learning to non-destructively track quality degradation over time, offering a promising step toward integrating AI-driven monitoring systems into the fresh food supply chain and reducing reliance on destructive testing methods.