Oracle bone heritage data augmentation based on two-stage decomposition GANs
Source abstract
Abstract
Oracle bone inscriptions(OBI) constitute a significant part of Chinese civilization, with many mysterious characters remaining undeciphered. Recent advances in deep learning have introduced novel approaches for OBI decipherment. However, OBI samples, often derived from tortoise shells and bones, are prone to corrosion and weathering, degrading dataset quality. This paper proposes an OBI augmentation model based on two-stage decomposition GANs, learning an unidirectional mapping to transform low-quality samples into high-quality ones. Unlike traditional data augmentation approaches that require pre-defined standard samples, our model utilizes a two-stage decomposition framework to augment OBI samples exclusively from existing data, ensuring enhanced realism and diversity, as measured by the Inception Score. Compared to existing augmentation techniques, our approach improves the Inception Score by 6.7% and reduces the Brisque Score by 30.2% on the public dataset. The augmented dataset quality improves recognition accuracy by 2.56%, demonstrating our approach’s effectiveness in real-world OBI decipherment.
Source: openalex_inverted_index · Quality: baseline
Research context
OpenAlex topics: Medical Imaging and Analysis · Medical Imaging Techniques and Applications · Image Processing and 3D Reconstruction
SciVX discovery tags are terminology-based. OpenAlex topics are source-provided classifications.
References (46)
Show source-provided references
1. Fu, X., Zhou, R., Yang, X. & Li, C. Detecting oracle bone inscriptions via pseudo-category labels. Herit. Sci. 12, 107 (2024).10.1186/s40494-024-01221-5 ↗
2. Anzhu, L. et al. Oracle-bone inscriptions and cultural memory. Front. Art. Res. 2, 63–73 (2020).
3. Li, J. et al. A comprehensive survey of oracle character recognition: challenges, benchmarks, and beyond. arXiv preprint arXiv:2411.11354 (2024).10.2139/ssrn.5081652 ↗
4. Guo, J., Wang, C., Roman-Rangel, E., Chao, H. & Rui, Y. Building hierarchical representations for oracle character and sketch recognition. IEEE Trans. Image Process. 25, 104–118 (2015).10.1109/TIP.2015.2500019 ↗
5. Han, W. et al. Self-supervised Learning of Orc-Bert Augmentor for Recognizing Few-Shot Oracle Characters. Asian Conference on Computer Vision, 652–668 (Cham, Springer International Publishing, 2020).10.1007/978-3-030-69544-6_39 ↗
6. Guan, H. et al. An open dataset for the evolution of oracle bone characters: Evobc. arXiv preprint arXiv:2401.12467 (2024).
7. 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 ↗
8. Huang, S., Wang, H., Liu, Y., Shi, X. & Jin, L. Obc306: A large-scale oracle bone character recognition dataset. In 2019 International Conference on Document Analysis and Recognition (ICDAR). 681–688 (IEEE, 2019).10.1109/ICDAR.2019.00114 ↗
9. Shi, D. et al. RCRN: Real-world Character Image Restoration Network via Skeleton Extraction. Proceedings of the 30th ACM International Conference on Multimedia, 1177–1185 (Association for Computing Machinery, 2022).10.1145/3503161.3548344 ↗
10. Diao, X. et al. Toward zero-shot character recognition: a gold standard dataset with radical-level annotations. Proceedings of the 31st ACM International Conference on Multimedia, 6869–6877 (2023).10.1145/3581783.3612201 ↗
11. Wang, M. & Deng, W. A dataset of oracle characters for benchmarking machine learning algorithms. Sci. Data 11, 87 (2024).10.1038/s41597-024-02933-w ↗
12. Wang, K., Yi, Y., Liu, J., Lu, L. & Song, Y. Multi-scene ancient chinese text recognition. Neurocomputing 377, 64–72 (2020).10.1016/j.neucom.2019.10.029 ↗
13. 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 ↗
14. Guo, Z. et al. An improved neural network model based on inception-v3 for oracle bone inscription character recognition. Sci. Program. 2022, 7490363 (2022).
15. Li, B. et al. Vit-based classification of oracle bones for periodization of the Shang dynasty. In International Conference on Algorithms, High Performance Computing, and Artificial Intelligence (AHPCAI 2023), vol. 12941, 575–579 (SPIE, 2023).10.1117/12.3011641 ↗
16. Ma, Y. et al. Stef: a swin transformer-based enhanced feature pyramid fusion model for dongba character detection. Herit. Sci. 12, 206 (2024).10.1186/s40494-024-01321-2 ↗
17. Zheng, Y., Chen, Y., Wang, X., Qi, D. & Yan, Y. Ancient chinese character recognition with improved swin-transformer and flexible data enhancement strategies. Sensors 24, 2182 (2024).10.3390/s24072182 ↗
18. Simonyan, K. & Zisserman, A. Very deep convolutional networks for large-scale image recognition. arXiv preprint arXiv:1409.1556 (2014).
19. Sermanet, P. et al. Overfeat: Integrated recognition, localization and detection using convolutional networks. 2nd International Conference on Learning Representations (ICLR 2014).
20. He, K., Zhang, X., Ren, S. & Sun, J. Deep residual learning for image recognition. In Proceedings of the IEEE conference on computer vision and pattern recognition, 770–778 (IEEE, 2016).10.1109/CVPR.2016.90 ↗
21. Dosovitskiy, A. et al. An image is worth 16x16 words: Transformers for image recognition at scale. arXiv preprint arXiv:2010.11929 (2020).
22. Liu, Z. et al. Swin transformer: Hierarchical vision transformer using shifted windows. In Proceedings of the IEEE/CVF International Conference on Computer Vision. 10012–10022 (IEEE, 2021).10.1109/ICCV48922.2021.00986 ↗
23. Shi, D. et al. CharFormer: A Glyph Fusion based Attentive Framework for High-precision Character Image Denoising. Proceedings of the 30th ACM International Conference on Multimedia, 1147–1155 (Association for Computing Machinery, 2022).10.1145/3503161.3548208 ↗
24. Cai, J. et al. TH-GAN: Generative Adversarial Network Based Transfer Learning for Historical Chinese Character Recognition. 2019 International Conference on Document Analysis and Recognition (ICDAR), 178–183 (IEEE Computer Society, 2019).10.1109/ICDAR.2019.00037 ↗
25. Li, J. et al. Diff-oracle: Diffusion model for oracle character generation with controllable styles and contents. arXiv preprint arXiv:2312.13631 (2023).
26. Wen, Q. et al. ZiGAN: Fine-grained Chinese Calligraphy Font Generation via a Few-shot Style Transfer Approach. Proceedings of the 29th ACM International Conference on Multimedia, 621–629 (ACM, 2021).10.1145/3474085.3475225 ↗
27. Wenjun, Z., Benpeng, S., Ruiqi, F., Xihua, P. & Shanxiong, C. Ea-gan: restoration of text in ancient chinese books based on an example attention generative adversarial network. Herit. Sci. 11, 1–13 (2023).10.1186/s40494-023-00882-y ↗
28. Chen, S., Yang, Y., Liu, X. & Zhu, S. Dual discriminator gan: Restoring ancient yi characters. Trans. Asian Low. Resour. Lang. Inf. Process. 21, 1–23 (2022).10.1145/3543851 ↗
29. Su, B., Liu, X., Gao, W., Yang, Y. & Chen, S. A restoration method using dual generate adversarial networks for chinese ancient characters. Vis. Inform. 6, 26–34 (2022).10.1016/j.visinf.2022.02.001 ↗
30. Huang, H. et al. AGTGAN: Unpaired Image Translation for Photographic Ancient Character Generation. Proceedings of the 30th ACM International Conference on Multimedia, 5456–5467 (Association for Computing Machinery, 2022).10.1145/3503161.3548338 ↗
31. 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 ↗
32. Wang, W. et al. Improving oracle bone characters recognition via a cyclegan-based data augmentation method. In International Conference on Neural Information Processing. 88–100 (Springer, 2022).10.1007/978-981-99-1645-0_8 ↗
33. Chang, X. et al. Sundial-GAN: A Cascade Generative Adversarial Networks Framework for Deciphering Oracle Bone Inscriptions. 30th ACM International Conference on Multimedia, MM 2022, 1195–1203 (Association for Computing Machinery, Inc, 2022).10.1145/3503161.3547925 ↗
34. Vaswani, A. et al. Attention is all you need. Adv. Neural Inf. Process. Syst. 30 https://arxiv.org/abs/1706.03762 (2017).
35. Gatys, L. A., Ecker, A. S. & Bethge, M. A. neural algorithm of artistic style. arXiv preprint arXiv:1508.06576 (2015).10.1167/16.12.326 ↗
36. Johnson, J., Alahi, A. & Fei-Fei, L. Perceptual losses for real-time style transfer and super-resolution. In Computer Vision–ECCV 2016: 14th European Conference, Amsterdam, The Netherlands, October 11–14, 2016, Proceedings, Part II 14, 694–711 (Springer, 2016).10.1007/978-3-319-46475-6_43 ↗
37. Wu, Q., Fan, C., Li, Y., Li, Y. & Hu, J. A novel perceptual loss function for single image super-resolution. Multimed. Tools Appl. 79, 21265–21278 (2020).10.1007/s11042-020-08878-7 ↗
38. Maas, A. L., Hannun, A. Y. & Ng, A. Y. Rectifier nonlinearities improve neural network acoustic models, In Proc. ICML, Vol. 30, 3 (Citeseer, 2013).
39. Goodfellow, I. J. et al. Generative adversarial nets. Adv. Neural Information Process. Syst 27. https://arxiv.org/abs/1406.2661 (2014).
40. Mescheder, L., Geiger, A. & Nowozin, S. Which training methods for gans do actually converge? In International Conference on Machine Learning. 3481–3490 (PMLR, 2018).
41. Salimans, T. et al. Improved techniques for training GANs. Adv. Neural Inf. Process. Syst. 29. https://arxiv.org/abs/1606.03498 (2016).
42. Heusel, M., Ramsauer, H., Unterthiner, T., Nessler, B. & Hochreiter, S. Gans trained by a two time-scale update rule converge to a local nash equilibrium. Adv. Neural Information Process. Syst. 30 https://arxiv.org/abs/1706.08500 (2017).
43. Mittal, A., Moorthy, A. K. & Bovik, A. C. No-reference image quality assessment in the spatial domain. IEEE Trans. Image Process. 21, 4695–4708 (2012).10.1109/TIP.2012.2214050 ↗
44. 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 ↗
45. Zhuo, S. & Zhang, J. Attention-based deformable convolutional network for chinese various dynasties character recognition. Expert Syst. Appl. 238, 121881 (2024).10.1016/j.eswa.2023.121881 ↗
46. Zhou, Y. et al. UFCNet: Unsupervised network based on Fourier transform and convolutional attention for oracle character recognition. In Proceedings of the 1st Workshop on Machine Learning for Ancient Languages (ML4AL 2024), 98–106 (2024).10.18653/v1/2024.ml4al-1.11 ↗