基于YOLO11和改进ResNet34的古籍印章识别模型研究

打开文本图片集

中图分类号：G256；TP391 文献标识码：A

Research on Ancient Book Seal Recognition Models Based on YOLO11 and Improved ResNet34

Yao ShunQu Yanling Long Huan Wang Qiuyun

Abstract Identifying seals inancient books isachallenging task.To address this problem，thispaperadopts atwo-stage approach forancientbook sealrecognition.First，YOLOl isemployed toachieve high-precisiondetectionofseal argets. Subsequently，an improved ResNet model integratingan ensemble self-attntion mechanism （Focus-ResNet34） is introduced torecognizeseal content.Inadition，an incremental learning strategy basedonuser fedback is applied， enabling the model tocontinuously learn and recognize new types of seals.Experimental results demonstrate that the proposedancient book sealrecognition modelcanaccurately identifyseals in imagesof ancient booksandoutput correspondingseal inscription information，significantly improving recognitionaccuracy.Themodel providesanew technicalapproach forancient book research and offersefectivesupport for therevitalization，utilization，and dissemination of ancient books.

Key Wordsancientbookseals;deep learning; neuralnetworks;objectdetection; YOLO11; ResNet34;attentionmechanism

0 引言

迹，了解书籍在不同历史时期所经历的变迁，从而更加深入地理解古籍的文化价值和历史意义。（剩余10373字）