3D Packaging Digital Reconstruction and Structural Similarity Analysis from 2D Drawings: An Industry-Engaged AI-Driven Syntegrative Education Project

The integration of artificial intelligence (AI) with real-world engineering challenges creates new opportunities for both technological innovation and educational transformation. However, existing approaches to packaging digitization often treat drawing understanding, 3D reconstruction, and similarity analysis as isolated tasks, limiting their effectiveness for design reuse. To address this gap, this paper proposes a unified AI-enabled framework that integrates large language model (LLM)-assisted drawing interpretation, parametric 3D reconstruction, and hybrid structural similarity analysis into a coherent pipeline. Unlike conventional methods, the framework combines semantic extraction, structure-aware modeling, and multi-level similarity reasoning to generate structured and machine-interpretable representations of legacy 2D packaging drawings. Specifically, structured information is extracted through LLM-based parsing with validation, followed by lightweight parametric 3D modeling. A hybrid similarity metric supports design retrieval, while an interactive interface enhances interpretability. A prototype demonstration shows the feasibility of the proposed framework in achieving reconstruction consistency and similarity discrimination. The framework is developed within a Syntegrative Education setting, an educational paradigm that integrates academic learning with industry-driven co-creation, highlighting its value for both industrial applications and practice-oriented learning. A student-centered evaluation further confirms measurable gains in AI competency and cross-disciplinary problem-solving skills among participating learners.