基于深度学习模型的中国当代红木圈椅智能识别方法探索  

Exploring Intelligent Recognition Methods for Contemporary Chinese Rosewood Round-Back Armchairs Based on Deep Learning Models

作　　者：张天星 冯雨[2] 晁垚 许世华 陈振益[1] 张凯锋 ZHANG Tianxing;FENG Yu;CHAO Yao;XU Shihua;CHEN Zhenyi;ZHANG Kaifeng(School of Art and Design,Wuyi University,Jiangmen,Guangdong,529020,China;School of Design and Art,Henan University of Technology,Zhengzhou,Henan,410000,China;College of Furnishings and Industrial Design,Nanjing Forestry University,Nanjing,Jiangsu,210037,China;Shanghai Electric Yantong Architectural Technology Group,Shanghai,200333,China)

机构地区：[1]五邑大学艺术与设计学院,广东江门529020 [2]河南工业大学设计艺术学院,河南郑州410000 [3]南京林业大学家居与工业设计学院,江苏南京210037 [4]上海电气研砼建筑科技集团,上海200333

出　　处：《家具与室内装饰》2025年第3期124-129,共6页Furniture & Interior Design

基　　金：2024年度国家社科基金艺术学项目(24BG151);2024年度校级本科教学质量与教学改革工程建设项目(JX2024051);2023年广东省高等教育教学改革类项目(GDJX2023022)。

摘　　要：深度学习作为人工智能最重要的研究方向,为中国传统家具的研究提供了新思路。以语义分割为切入点,从样本收集、文件标注、模型的训练、评估与识别验证方面探索中国当代红木圈椅的智能识别方法。在样本收集方面,选择具有代表性的明清经典款圈椅作为训练与测试的样张;在标注方面,利用Labelme对样张进行标注,标注的数据被划分为训练集与测试集;在模型训练方面,以KNet作为网络架构,对训练样张进行机器学习,得出精度较高的模型;在评估方面,通过测试集上评估指标的识别准确度评判模型的性能;在验证方面,以当代红木圈椅为识别对象,借助识别结果对模型的识别能力进行验证。在数字化飞速普及的背景下,深度学习技术的引入能为传统红木家具行业的发展与相关研究提供一定的参考。As one of the most important research directions in artificial intelligence,deep learning offers new perspectives for the study of traditional Chinese furniture.Taking semantic segmentation as the entry point,this study explores intelligent recognition methods for contemporary Chinese rosewood round-back armchairs from the aspects of sample collection,document annotation,model training,evaluation,and recognition verification.For sample collection,representative classic styles of round-back armchairs from the Ming and Qing dynasties are selected as training and testing samples.In terms of annotation,Labelme is used to label the samples,and the annotated data are divided into training and testing sets.For model training,the KNet network architecture is employed to conduct machine learning on the training samples,resulting in a model with relatively high accuracy.In the evaluation phase,the model’s performance is assessed based on the recognition accuracy of the evaluation metrics applied to the test set.For validation,contemporary rosewood round-back armchairs are used as recognition targets and the recognition results are used to verify the model’s recognition capability.In the context of the rapid spread of digitalization,introducing deep learning technology can offer references for the development of the traditional rosewood furniture industry and related researches.

关 键 词：深度学习 语义分割 KNet 当代红木圈椅 智能识别 

分 类 号：TS664.1[轻工技术与工程]