Artificial intelligence–based detection of defects in meat packaging films: a review

The safety and quality of meat products are paramount in the food industry, with packaging films playing a crucial role in preserving freshness and preventing contamination. However, defects in packaging films can compromise product integrity, leading to food spoilage, economic losses, and potential health hazards. Packaging films can develop micro-tears, seal failures, pinholes, delamination, and structural weaknesses. Such defects may impair the protective performance of the packaging, creating pathways for unwanted gas exchange, moisture gain or loss, and incidental microbial contact. These changes can reduce product quality, promote early spoilage, and increase the likelihood of contamination that could threaten consumer safety. Conventional detection relies on visual inspection, manual pressure testing, dye infiltration, and basic leak assessments; however, fine micro-defects and incipient seal failures often remain undetected, enabling compromised meat-packaging films to pass routine inspection. Artificial intelligence (AI) has emerged as a transformative tool for addressing these challenges by integrating advanced imaging, sensing, and computational analysis. AI-enabled technologies, including computer vision, hyperspectral imaging, ultrasound, X-ray imaging, thermography, and intelligent packaging, enable rapid, non-destructive, and highly accurate identification of defects previously undetectable by traditional methods. This review synthesizes recent scientific advances in AI-based detection of defects in meat packaging films, with particular emphasis on packaging integrity, sealing defects, multilayer-film failures, and complementary monitoring of packaged meat quality, highlighting deep learning architectures, data analytics, and intelligent sensor systems. Key challenges, including limited annotated datasets, class imbalance, domain shift across heterogeneous production environments, explainability requirements in regulated food systems, and the evolving characteristics of biodegradable packaging materials, are critically examined. Emerging trends emphasize multimodal sensor fusion, edge computing, self-supervised learning, and standardized validation frameworks to improve scalability, robustness, and regulatory acceptance. The review evaluates the strengths and limitations of AI-driven inspection approaches, links packaging-film defect detection with intra-meat defect identification and outlines strategic directions for industrial implementation. It concludes with forward-looking perspectives aimed at improving the accuracy, reliability, and efficiency of defective detection systems in meat packaging applications.