plSSN : 0374-8111 | elSSN : 2287-8815
plSSN : 0374-8111 | elSSN : 2287-8815
plSSN : 0374-8111elSSN : 2287-8815
Artificial Intelligence-Enabled Sensing for Seafood Quality and SafetyAssessment
심은지·현지민·김재일·류보미*
국립부경대학교 수산과학대학 식품영양학전공
Seafood is highly susceptible to quality deterioration and safety risks during post-harvest handling and storage, creating a growing demand for rapid, non-destructive, and reliable assessment technologies. This review critically examines recent advances in sensing technologies combined with artificial intelligence (AI) for seafood quality and safety, with emphasis on computer vision, spectroscopy and hyperspectral imaging, electronic sensing and biosensing, and multimodal data fusion. Computer vision is particularly useful for quantifying visible quality changes, whereas spectroscopic approaches provide information on physicochemical and biochemical alterations. Electronic sensing and biosensing enable the detection of spoilage-related patterns and target-specific safety hazards. Recent studies have increasingly integrated multiple sensing modalities through data-, feature-, and decision-level fusion to capture complementary aspects of seafood deterioration. However, improved model accuracy does not necessarily translate into robust real-world performance. Major limitations remain with respect to dataset size, biological replication, target definition, external validation, instrument dependency, explainability, and environmental variability. Future research should therefore prioritize generalizable and interpretable models, rational sensor combinations, and validation across species, batches, instruments, and processing environments. Sensing combined with AI has strong potential to support practical seafood quality control and safety screening once its performance has been demonstrated beyond laboratory-scale proof of concept.
Artificial intelligence, Multimodal data fusion, Seafood quality, Seafood safety, Sensing technology