Enhancing consumer perception and market value of traditional fermented rice cakes via AI-driven cultural packaging: A hybrid framework.

The modernization of traditional food products faces a critical challenge: generic packaging often fails to convey historical value, thereby reducing consumer purchase intent. Using Hemudu Fermented Rice Cake as a case study, this study proposes a closed-loop packaging development framework to strengthen brand distinctiveness and improve consumer perception. A "closed-set fidelity" paradigm is developed by integrating Shape Grammar, CRITIC, and LoRA (Low-Rank Adaptation). In contrast to unconstrained generative AI, which may produce patterns with visual inconsistency or semantic inaccuracy, the proposed method applies Shape Grammar to extract core morphological rules and then uses CRITIC to screen these rules and construct a high-quality dataset. Based on this dataset, the LoRA model generates high-fidelity packaging patterns that are suitable for industrial printing. Finally, a multi-criteria sensory evaluation based on F-TOPSIS was conducted with 50 panelists, including consumers and experts. The results show that the optimized packaging design (Scheme S8) performed significantly better than the baseline schemes in "Purchase Intent Stimulation" and "Food Product Compatibility." This study provides a practical approach to enhancing the cultural expression and market value of traditional food products and demonstrates that AI-assisted design can improve the clarity of quality cues in food packaging.