HUNet: hierarchical universal network for multi-type ancient Chinese character recognition
Source abstract
Abstract
Abstract Ancient Chinese characters preserved in ancient books, calligraphy paintings, stones, and bronzes rubbings exhibit diverse morphological structures distinct from modern scripts, requiring specialized paleographic expertise for accurate recognition. However, few experts can recognize multiple character types simultaneously. This study constructed the Multi-Type Ancient Chinese Character Recognition (MTACCR) dataset and proposed HUNet, a Hierarchical Universal Network that employs a parameter-sharing architecture to achieve efficient recognition of diverse ancient character types through multi-stage feature extraction and fusion. The experimental results show that compared to other efficient models evaluated, the HUNet series achieves higher recognition performance with similar parameter counts (or maintains lower parameter counts at equivalent recognition performance) while sustaining high computational throughput. In addition, after training on MTACCR, HUNet’s Top-1 accuracy on the MACR test set improved by 9.64%. Therefore, HUNet can be easily deployed on different devices and platforms for efficiently recognizing multiple types of ancient Chinese characters.
Source: crossref · Quality: baseline
Research context
OpenAlex topics: Handwritten Text Recognition Techniques · Image Processing and 3D Reconstruction · Currency Recognition and Detection
SciVX discovery tags are terminology-based. OpenAlex topics are source-provided classifications.
References (37)
Show source-provided references
1. Zhang, M. The ‘Chinese wisdom’ in Chinese characters: an analysis of the evolution of Chinese characters and their contemporary value. Hanzi Cult. 04, 96–98 (2023).
2. Liu, J. On the fonts and types of Chinese characters. J. Shaoxing Univ. (Philos. Soc. Sci.) 01, 77–82 (2007).
3. Zhang, H. A critical analysis of the development origins of Chinese character fonts. J. Inn. Mong. Norm. Univ. (Philos. Soc. Sci.) 03, 92–94 (2007).
4. Zhu, L. A discussion on the evolution of ancient Chinese character forms. J. Mudanjiang Norm. Univ. (Philos. Soc. Sci.) 01, 95–96 (2015).
5. Ma, X. A study on the evolution and development of Chinese character forms. Hanzi Cult. 11, 1–3 (2022).10.3366/cult.2022.0252 ↗
6. Yang, Q. A review of Qiu Xigui’s “Outline of philology”: also discussing the evolution of Chinese character forms and simplified characters. Hanzi Cult. 24, 4–5 (2022).
7. Li, B. et al. HWOBC-A handwriting oracle bone character recognition database. J. Phys. Conf. Ser. 1651, 012050 (2020).10.1088/1742-6596/1651/1/012050 ↗
8. Wang, M., Deng, W. & Liu, C.-L. Unsupervised structure-texture separation network for oracle character recognition. IEEE Trans. Image Process. 31, 3137–3150 (2022).10.1109/TIP.2022.3165989 ↗
9. Li, J. et al. Towards better long-tailed oracle character recognition with adversarial data augmentation. Pattern Recognit. 140, 109534 (2023).10.1016/j.patcog.2023.109534 ↗
10. Wang, M., Deng, W. & Su, S. Oracle character recognition using unsupervised discriminative consistency network. Pattern Recognit. 148, 110180 (2024).10.1016/j.patcog.2023.110180 ↗
11. Li, S., Zhou, C. & Wang, K. WA-Net: Wavelet integrated attention network for silk and bamboo character recognition. Eng. Appl. Artif. Intell. 140, 109674 (2025).10.1016/j.engappai.2024.109674 ↗
12. Xu, Y. et al. CASIA-AHCDB: a large-scale Chinese ancient handwritten characters database. In Proc. 2019 International Conference on Document Analysis and Recognition (ICDAR), 793-798 (2019).10.1109/ICDAR.2019.00132 ↗
13. Miao, Y., Liang, L., Ji, Y., Li, Z. & Li, G. Research on Chinese ancient characters image recognition method based on adaptive receptive field. Soft Comput. 26, 8273–8282 (2022).10.1007/s00500-022-07270-x ↗
14. Sun, J., Li, P. & Wu, X. Handwritten ancient Chinese character recognition algorithm based on improved Inception-ResNet and attention mechanism. In Proc. 2022 IEEE 2nd International Conference on Software Engineering and Artificial Intelligence (SEAI), 31-35 (2022).10.1109/SEAI55746.2022.9832356 ↗
15. Wang, K., Yi, Y., Liu, J., Lu, L. & Song, Y. Multi-scene ancient Chinese text recognition. Neurocomput 377, 64–72 (2020).10.1016/j.neucom.2019.10.029 ↗
16. Wang, K., Yi, Y., Tang, Z. & Peng, J. Multi-scene ancient Chinese text recognition with deep coupled alignments. Appl. Soft Comput. 108, 107475 (2021).10.1016/j.asoc.2021.107475 ↗
17. Zhang, J. et al. Rethinking mobile block for efficient attention-based models. In Proc. 2023 IEEE/CVF International Conference on Computer Vision (ICCV), 1389–1400 (2023).10.1109/ICCV51070.2023.00134 ↗
18. Siahaan, D., Sutramiani, N. P., Suciati, N., Duija, I. N. & Darma, I. W. A. S. DeepLontar dataset for handwritten Balinese character detection and syllable recognition on Lontar manuscript. Sci. Data 9, 761 (2022).10.1038/s41597-022-01867-5 ↗
19. Swindall, M. I. et al. Exploring learning approaches for ancient Greek character recognition with citizen science data. In Proc. 2021 IEEE 17th International Conference on eScience (eScience), 128–137 (2021).10.1109/eScience51609.2021.00023 ↗
20. Chen, P., Liu, S., Zhao, H. & Jia, J. GridMask data augmentation. Preprint at https://arxiv.org/abs/2001.04086 (2020).
21. Chen, J. et al. Run, don’t walk: chasing higher FLOPS for faster neural networks. In Proc. 2023 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) 12021–12031 (2023).10.1109/CVPR52729.2023.01157 ↗
22. Wang, Q. et al. ECA-Net: Efficient channel attention for deep convolutional neural networks. In Proc. 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 11531–11539 (2020).10.1109/CVPR42600.2020.01155 ↗
23. Wang, G., Zhao, Y., Tang, C., Luo, C. & Zeng, W. When shift operation meets vision transformer: an extremely simple alternative to attention mechanism. Proc. AAAI Conf. Artif. Intell. 36, 2423–2430 (2022).
24. Cui, J., Liu, S., Tian, Z., Zhong, Z. & Jia, J. ResLT: residual learning for long-tailed recognition. IEEE Trans. Pattern Anal. Mach. Intell. 45, 3695–3706 (2023).
25. Li, J. et al. Next-vit: Next generation vision transformer for efficient deployment in realistic industrial scenarios. Preprint at https://arxiv.org/abs/2207.05501 (2022).
26. Liu, Z. et al. Open long-tailed recognition in a dynamic world. IEEE Trans. Pattern Anal. Mach. Intell. 46, 1836–1851 (2022).10.1109/TPAMI.2022.3200091 ↗
27. Zhang, Y., Kang, B., Hooi, B., Yan, S. & Feng, J. Deep long-tailed learning: a survey. IEEE Trans. Pattern Anal. Mach. Intell. 45, 10795–10816 (2023).10.1109/TPAMI.2023.3268118 ↗
28. Howard, A. G. et al. Searching for MobileNetV3. In Proc. 2019 IEEE/CVF International Conference on Computer Vision (ICCV), 1314–1324 (2019).10.1109/ICCV.2019.00140 ↗
29. Hou, Q., Zhou, D. & Feng, J. Coordinate attention for efficient mobile network design. In Proc. 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 13708–13717 (2021).10.1109/CVPR46437.2021.01350 ↗
30. Guan, H. et al. An open dataset for the evolution of oracle bone characters: Evobc. Preprint at https://arxiv.org/abs/2401.12467 (2024).
31. Wang, P. et al. An open dataset for oracle bone character recognition and decipherment. Sci. Data 11, 976 (2024).10.1038/s41597-024-03807-x ↗
32. Liu, Y. et al. DeepJiandu dataset for character detection and recognition on Jiandu manuscript. Sci. Data 12, 398 (2025).10.1038/s41597-025-04716-3 ↗
33. McInnes, L., Healy, J., Saul, N. & Grossberger, L. UMAP: Uniform Manifold Approximation and Projection. J. Open Source Softw. 3, 861 (2018).10.21105/joss.00861 ↗
34. Sandler, M., Howard, A. G., Zhu, M., Zhmoginov, A. & Chen, L.-C. MobileNetV2: Inverted residuals and linear bottlenecks. In Proc. 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 4510–4520 (2018).10.1109/CVPR.2018.00474 ↗
35. Tang, Y. et al. GhostNetv2: Enhance Cheap Operation with Long-Range Attention. Adv. Neural Inf. Process Syst. 35, 9969–9982 (2022).
36. Ma, N., Zhang, X., Zheng, H.-T. & Sun, J. ShuffleNet V2: Practical guidelines for efficient CNN architecture design. Comput. Vision–ECCV 2018, 122–138 (2018).10.1007/978-3-030-01264-9_8 ↗
37. Liu, X. et al. EfficientViT: Memory efficient vision transformer with cascaded group attention. In Proc. 2023 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 14420–14430 (2023).10.1109/CVPR52729.2023.01386 ↗