Integrating Deep Learning and Machine Learning for Ceramic Artifact Classification and Market Value Prediction

This study proposes an intelligent framework for the automated classification and valuation of ceramic artifacts, integrating deep learning and machine learning techniques. An improved YOLOv11 model was constructed to identify key ceramic attributes such as decorative patterns, shapes, and craftsmanship styles. The model achieved a mean Average Precision (mAP@50) of 70.0% and a recall of 91.0%, demonstrating strong capability in detecting complex visual features. Based on the extracted visual attributes, a Random Forest classifier was employed to predict price categories using multi-source auction data, achieving a test accuracy of 99.52%. Feature importance analysis further revealed manufacturing techniques and shape as key predictors of market value. The integrated framework effectively combines visual feature extraction and market-informed valuation, providing a scalable solution for intelligent ceramic appraisal and digital heritage curation. This approach supports both expert and non-expert applications, laying a foundation for future development of intelligent cultural heritage management systems.本研究提出一种整合深度学习与机器学习技术的陶瓷文物自动分类及估值智能框架。构建改进的YOLOv11模型以识别装饰图案、器形及工艺风格等关键陶瓷属性。该模型的平均精度均值（mean Average Precision，mAP@50）达70.0%，召回率为91.0%，展现出对复杂视觉特征的强大检测能力。基于提取的视觉属性，研究采用随机森林（Random Forest）分类器，结合多源拍卖数据预测价格类别，测试准确率达99.52%。特征重要性分析进一步揭示，制造工艺与器形是市场价值的关键预测因素。该整合框架有效结合视觉特征提取与市场导向估值，为智能陶瓷鉴定及数字遗产管理提供可扩展解决方案。此方法支持专家与非专家应用场景，为智能文化遗产管理系统的未来发展奠定基础。