Research on Computer Vision Teaching of Mongolian Silver Jewelry Making Techniques

• Jun, Xi ^[1] ; Sukhbayar , Battsooj ^[2]
  1. [1] Mongolian State University of Education. Mongolian
  2. [2] School of Fine Arts and Technology, Mongolian State University of Education，Mongolian
• Localización: Metaverse Basic and Applied Research, ISSN-e 2953-4577, Vol. 4, Nº. 0, 2025 (Ejemplar dedicado a: Metaverse Basic and Applied Research)
• Idioma: inglés
• Enlaces
  • Texto completo (pdf)
• Resumen

  • Introduction：Jewelry making is an integral part of Mongolian craftsmanship, reflecting the country's rich cultural heritage and artistic traditions. However, traditional methods often lack mechanisms for real-time feedback, limiting learner’s ability to detect and correct production defects. To address this issue, this research introduces computer vision-based teaching framework that integrates Deep learning (DL) for analyzing handcrafted silver jewelry, enhancing learning outcomes and preserving traditional artistry.

    Method：The research utilizes the open-source Mongolian Silver Jewelry Defect Dataset (MSJDD) from Kaggle, comprising 1,050 samples of handcrafted jewelry.  A Scalable Transient Search-tuned Multi-Cascaded Convolutional Neural Network (STS-MCCNN) model was developed to classify jewelry pieces, identify surface defects, and evaluate craftsmanship quality. The dataset was preprocessed using Min-max scaling to reduce noise, while Principal Component Analysis (PCA) was applied to improve feature extraction. The AI-driven interface enables users to input jewelry characteristics, receive automatic defect analysis, and visualize analytical heat maps highlighting critical defect zones, supported by an adaptive feedback mechanism for skill refinement.

    Results：Experimental evaluations revealed that the proposed STS-MCCNN model achieved 98.5% accuracy, 98.36% recall, and 98.25% F1-score, supporting its high reliability in defect detection and craftsmanship evaluation. Moreover, the integration of real-time feedback significantly improved learner engagement and precision in jewelry-making techniques.

    Conclusions：This research demonstrates how combining traditional artistry with AI technologies can preserve and modernize Mongolian jewelry-making practices. The proposed STS-MCCNN method enhances both learning and cultural continuity, offering a sustainable pathway for advancing artisan education and safeguarding intangible cultural heritage.