請用此 Handle URI 來引用此文件:
http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/84429
完整後設資料紀錄
DC 欄位 | 值 | 語言 |
---|---|---|
dc.contributor.advisor | 黃乾綱(Chien-Kang Huang) | |
dc.contributor.author | Ting Wang | en |
dc.contributor.author | 王婷 | zh_TW |
dc.date.accessioned | 2023-03-19T22:11:19Z | - |
dc.date.copyright | 2022-09-30 | |
dc.date.issued | 2022 | |
dc.date.submitted | 2022-09-26 | |
dc.identifier.citation | [1] N. Arica and F. T. Yarman-Vural, 'An overview of character recognition focused on off-line handwriting,' IEEE Transactions on Systems, Man and Cybernetics, Part C (Applications and Reviews), vol. 31, no. 2, pp. 216-233, 2001-05-01 2001, doi: 10.1109/5326.941845. [2] X.-Y. Zhang, Y. Bengio, and C.-L. Liu, 'Online and offline handwritten Chinese character recognition: A comprehensive study and new benchmark,' Pattern Recognition, vol. 61, pp. 348-360, 2017-01-01 2017, doi: 10.1016/j.patcog.2016.08.005. [3] C.-L. Liu, F. Yin, D.-H. Wang, and Q.-F. Wang, 'CASIA Online and Offline Chinese Handwriting Databases,' in 2011 International Conference on Document Analysis and Recognition, 2011-09-01 2011: IEEE, doi: 10.1109/icdar.2011.17. [4] R. Dai, C. Liu, and B. Xiao, 'Chinese character recognition: history, status and prospects,' Frontiers of Computer Science in China, vol. 1, no. 2, pp. 126-136, 2007-05-01 2007, doi: 10.1007/s11704-007-0012-5. [5] F. Kimura, K. Takashina, S. Tsuruoka, and Y. Miyake, 'Modified Quadratic Discriminant Functions and the Application to Chinese Character Recognition,' IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. PAMI-9, no. 1, pp. 149-153, 1987-01-01 1987, doi: 10.1109/tpami.1987.4767881. [6] Z. Zhong, L. Jin, and Z. Xie, 'High performance offline handwritten Chinese character recognition using GoogLeNet and directional feature maps,' in 2015 13th International Conference on Document Analysis and Recognition (ICDAR), 2015-08-01 2015: IEEE, doi: 10.1109/icdar.2015.7333881. [7] D. Ciresan and U. Meier, 'Multi-Column Deep Neural Networks for offline handwritten Chinese character classification,' in 2015 International Joint Conference on Neural Networks (IJCNN), 2015-07-01 2015: IEEE, doi: 10.1109/ijcnn.2015.7280516. [8] M. Z. Alom et al., 'A State-of-the-Art Survey on Deep Learning Theory and Architectures,' Electronics, vol. 8, no. 3, p. 292, 2019-03-05 2019, doi: 10.3390/electronics8030292. [9] A. Torralba and A. A. Efros, 'Unbiased look at dataset bias,' in CVPR 2011, 2011-06-01 2011: IEEE, doi: 10.1109/cvpr.2011.5995347. [10] W. J. Scheirer, L. P. Jain, and T. E. Boult, 'Probability Models for Open Set Recognition,' IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 36, no. 11, pp. 2317-2324, 2014-11-01 2014, doi: 10.1109/tpami.2014.2321392. [11] P. R. Mendes Júnior et al., 'Nearest neighbors distance ratio open-set classifier,' Machine Learning, vol. 106, no. 3, pp. 359-386, 2017-03-01 2017, doi: 10.1007/s10994-016-5610-8. [12] C. Geng, S.-J. Huang, and S. Chen, 'Recent Advances in Open Set Recognition: A Survey,' IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 43, no. 10, pp. 3614-3631, 2021-10-01 2021, doi: 10.1109/tpami.2020.2981604. [13] Z. Li, Q. Wu, Y. Xiao, M. Jin, and H. Lu, 'Deep Matching Network for Handwritten Chinese Character Recognition,' Pattern Recognition, vol. 107, p.107471, 2020/11/01/ 2020, doi: https://doi.org/10.1016/j.patcog.2020.107471. [14] Ministry of Education Republic of China (Taiwan), '教育部4808 個常用字.' [Online]. Available: https://language.moe.gov.tw/001/Upload/Files/site_content/download/mandr/ 教育部4808 個常用字.pdf. [15] Ministry of Education of the People's Republic of China, 'Table of General Standard Chinese Characters,' 2013 2013. [Online]. Available: http://www.moe.gov.cn/jyb_sjzl/ziliao/A19/201306/t20130601_186002.html. [16] Y. Zhang, S. Liang, S. Nie, W. Liu, and S. Peng, 'Robust offline handwritten character recognition through exploring writer-independent features under the guidance of printed data,' Pattern Recognition Letters, vol. 106, pp. 20-26, 2018-04-01 2018, doi: 10.1016/j.patrec.2018.02.006. [17] M. Wang and W. Deng, 'Deep visual domain adaptation: A survey,' Neurocomputing, vol. 312, pp. 135-153, 2018-10-01 2018, doi: 10.1016/j.neucom.2018.05.083. [18] P. Melnyk, Z. You, and K. Li, 'A high-performance CNN method for offline handwritten Chinese character recognition and visualization,' Soft Computing, vol. 24, no. 11, pp. 7977-7987, 2020-06-01 2020, doi: 10.1007/s00500-019-04083-3. [19] C. Szegedy et al., 'Going deeper with convolutions,' in 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2015-06-01 2015: IEEE, doi: 10.1109/cvpr.2015.7298594. [20] C. Luo, Y. Zhu, L. Jin, and Y. Wang, 'Learn to Augment: Joint Data Augmentation and Network Optimization for Text Recognition,' in 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2020-06-01 2020: IEEE, doi: 10.1109/cvpr42600.2020.01376. [21] T. Hayashi, K. Gyohten, H. Ohki, and T. Takami, 'A Study of Data Augmentation for Handwritten Character Recognition using Deep Learning,' in 2018 16th International Conference on Frontiers in Handwriting Recognition (ICFHR), 2018-08-01 2018: IEEE, doi: 10.1109/icfhr-2018.2018.00102. [22] B. Chang, Q. Zhang, S. Pan, and L. Meng, 'Generating Handwritten Chinese Characters Using CycleGAN,' in 2018 IEEE Winter Conference on Applications of Computer Vision (WACV), 2018-03-01 2018: IEEE, doi: 10.1109/wacv.2018.00028. [23] J. Zeng, Q. Chen, Y. Liu, M. Wang, and Y. Yao, 'StrokeGAN: Reducing Mode Collapse in Chinese Font Generation via Stroke Encoding,' Proceedings of the AAAI Conference on Artificial Intelligence, vol. 35, no. 4, pp. 3270-3277, 2021-05-18 2021, doi: 10.1609/aaai.v35i4.16438. [24] Z. Cao, J. Lu, S. Cui, and C. Zhang, 'Zero-shot Handwritten Chinese Character Recognition with hierarchical decomposition embedding,' Pattern Recognition, vol. 107, p. 107488, 2020-11-01 2020, doi: 10.1016/j.patcog.2020.107488. [25] W. Wang, J. Zhang, J. Du, Z.-R. Wang, and Y. Zhu, 'DenseRAN for Offline Handwritten Chinese Character Recognition,' in 2018 16th International Conference on Frontiers in Handwriting Recognition (ICFHR), 2018-08-01 2018: IEEE, doi: 10.1109/icfhr-2018.2018.00027. [26] T. Wang, Z. Xie, Z. Li, L. Jin, and X. Chen, 'Radical aggregation network for fewshot offline handwritten Chinese character recognition,' Pattern Recognition Letters, vol. 125, pp. 821-827, 2019, doi: 10.1016/j.patrec.2019.08.005. [27] X. Ao, X.-Y. Zhang, H.-M. Yang, F. Yin, and C.-L. Liu, 'Cross-Modal Prototype Learning for Zero-Shot Handwriting Recognition,' in 2019 International 41 Conference on Document Analysis and Recognition (ICDAR), 2019-09-01 2019: IEEE, doi: 10.1109/icdar.2019.00100. [28] S. Ioffe and C. Szegedy, 'Batch normalization: Accelerating deep network training by reducing internal covariate shift,' in International conference on machine learning, 2015: PMLR, pp. 448-456. [29] R. Hou, H. Chang, B. Ma, S. Shan, and X. Chen, 'Cross Attention Network for Few-shot Classification,' presented at the Advances in Neural Information Processing Systems 32: Annual Conference on Neural Information Processing Systems 2019, NeurIPS 2019, December 8-14, 2019, Vancouver, BC, Canada, 2019, 2019. [Online]. Available: https://proceedings.neurips.cc/paper/2019/hash/01894d6f048493d2cacde3c579c315a3-Abstract.html. [30] F. Yin, Q.-F. Wang, X.-Y. Zhang, and C.-L. Liu, 'ICDAR 2013 Chinese Handwriting Recognition Competition,' in 2013 12th International Conference on Document Analysis and Recognition, 2013-08-01 2013: IEEE, doi: 10.1109/icdar.2013.218. [31] Y. Xu, F. Yin, D.-H. Wang, X.-Y. Zhang, Z. Zhang, and C.-L. Liu, 'CASIAAHCDB: A Large-Scale Chinese Ancient Handwritten Characters Database,' in 2019 International Conference on Document Analysis and Recognition (ICDAR), 2019-09-01 2019: IEEE, doi: 10.1109/icdar.2019.00132. [32] ' 玉山人工智慧挑戰賽 2021 夏季賽.' [Online]. Available: https://tbrain.trendmicro.com.tw/Competitions/Details/14. [33] K. He, X. Zhang, S. Ren, and J. Sun, 'Delving Deep into Rectifiers: Surpassing Human-Level Performance on ImageNet Classification,' in 2015 IEEE International Conference on Computer Vision (ICCV), 2015-12-01 2015: IEEE, doi: 10.1109/iccv.2015.123. [34] L. Chen, S. Wang, W. Fan, J. Sun, and S. Naoi, 'Beyond human recognition: A CNN-based framework for handwritten character recognition,' in 2015 3rd IAPR Asian Conference on Pattern Recognition (ACPR), 2015-11-01 2015: IEEE, doi: 10.1109/acpr.2015.7486592. [35] X. Yang, D. He, Z. Zhou, D. Kifer, and C. L. Giles, 'Improving Offline Handwritten Chinese Character Recognition by Iterative Refinement,' in 2017 14th IAPR International Conference on Document Analysis and Recognition (ICDAR), 2017-11-01 2017: IEEE, doi: 10.1109/icdar.2017.11. [36] J. Ponce et al., 'Dataset Issues in Object Recognition,' in Toward Category-Level Object Recognition, J. Ponce, M. Hebert, C. Schmid, and A. Zisserman Eds., (Lecture Notes in Computer Science. Berlin, Heidelberg: Springer Berlin Heidelberg, 2006, ch. Chapter 2, pp. 29-48. [37] R. Taori, A. Dave, V. Shankar, N. Carlini, B. Recht, and L. Schmidt, 'Measuring Robustness to Natural Distribution Shifts in Image Classification,' presented at the Advances in Neural Information Processing Systems, 2020, 2020. [Online]. Available: https://proceedings.neurips.cc/paper/2020/file/d8330f857a17c53d217014ee776bfd50-Paper.pdf. [38] Y.-K. Zhang, H. Zhang, Y.-G. Liu, Q. Yang, and C.-L. Liu, 'Oracle Character Recognition by Nearest Neighbor Classification with Deep Metric Learning,' in 2019 International Conference on Document Analysis and Recognition (ICDAR), 2019-09-01 2019: IEEE, doi: 10.1109/icdar.2019.00057. [39] X. Fu, Z. Yang, Z. Zeng, Y. Zhang, and Q. Zhou, 'Improvement of Oracle Bone Inscription Recognition Accuracy: A Deep Learning Perspective,' ISPRS International Journal of Geo-Information, vol. 11, no. 1, p. 45, 2022-01-09 2022, doi: 10.3390/ijgi11010045. [40] R. Sears, 'Chinese etymology 字源,' Chinese Etymology 字源. [Online]. Available: https://hanziyuan.net/. [41] Academia Sinica, 'Xiaoxuetang Jiaguwen.' [Online]. Available: https://xiaoxue.iis.sinica.edu.tw/jiaguwen. [42] J. Guo, C. Wang, E. Roman-Rangel, H. Chao, and Y. Rui, 'Building Hierarchical Representations for Oracle Character and Sketch Recognition,' IEEE Transactions on Image Processing, vol. 25, no. 1, pp. 104-118, 2016-01-01 2016, doi: 10.1109/tip.2015.2500019. [43] K. Simonyan and A. Zisserman, 'Very Deep Convolutional Networks for Large-Scale Image Recognition,' 2014 2014, doi: 10.48550/ARXIV.1409.1556. [44] K. He, X. Zhang, S. Ren, and J. Sun, 'Deep Residual Learning for Image Recognition,' in 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2016-06-01 2016: IEEE, doi: 10.1109/cvpr.2016.90. [45] C. Sitawarin and D. Wagner, 'On the Robustness of Deep K-Nearest Neighbors,' in 2019 IEEE Security and Privacy Workshops (SPW), 2019-05-01 2019: IEEE, doi: 10.1109/spw.2019.00014. | |
dc.identifier.uri | http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/84429 | - |
dc.description.abstract | 近來的基於深度神經網路的模型在中文手寫辨識已高於人類辨識率。然而訓練集與真實環境的特徵分佈和類別分佈存在差距,當這類模型面對與訓練集存在特徵差異的資料準確率會下降,且無法直接用於辨識未學過的類別。因此本研究的目的是提出一個能夠在不微調或不重新訓練的情況下,模型能夠辨識不在訓練集內的類別,且對特徵變化的敏感度下降。 根據本研究的目的,我們透過訓練模型比較手寫字與印刷字相似性的方式,提出一個基於偽孿生網路架構的模型PSN-GC,透過給予新類別的印刷字範本,即可辨識不在訓練集中的類別。我們的方法相較過去研究提升了準確率,並降低記憶體用量與計算量。 實驗使用多種測試集對PSN-GC做全面的評估,測試條件可被歸類為閉集與開集。為了更進一步測試PSN-GC的極限,我們也使用甲骨文作為訓練集與測試集,因甲骨文的筆畫變化較現代手寫中文更高。以上實驗顯示我們的模型略遜於專精於閉集條件,也就是對已知類別最佳化的方法;但是與開集方法相比,我們的模型得到更高的準確率,且對特徵敏感度較低。 | zh_TW |
dc.description.abstract | Recently, deep neural network-based models have achieved higher performance than humans in handwritten Chinese character recognition. However, the feature distribution and label distribution of real-world data are different from training sets. The recognition rates will drop when this type of model is evaluated on real-world data. Also, the models can not recognize unlearned categories without retraining or finetuning. Therefore, this study aims at proposing a model that can be applied to open-set and is less sensitive to feature changes. According to the purpose of this research, by training the model to compare the similarity between handwritten and printed characters, we propose a model PSN-GC based on the pseudo-Siamese network architecture. Our method improves accuracy and consumes less memory usage and computation than previous studies. The experiments use multiple testing sets to conduct a comprehensive evaluation of PSN-GC, including closed-set conditions and open-set conditions. In order to further test the limit of PSN-GC, we also use oracle bone inscriptions as the training set and testing set due to the stroke variation of oracle bone script higher than modern handwritten Chinese characters. Though our model is less accurate than the models optimized to learned categories under closed-set conditions, our model achieves higher accuracy and is less sensitive to feature changes under open-set conditions. | en |
dc.description.provenance | Made available in DSpace on 2023-03-19T22:11:19Z (GMT). No. of bitstreams: 1 U0001-1609202219070900.pdf: 1970337 bytes, checksum: 5fe4b2b00e38d49b07090c9f5544be2e (MD5) Previous issue date: 2022 | en |
dc.description.tableofcontents | 誌謝 ii 中文摘要 iii ABSTRACT iv CONTENTS v LIST OF FIGURES viii LIST OF TABLES x Chapter 1 Introduction 1 1.1 Motivation 1 1.2 Main Contribution 2 1.3 Thesis structure 3 Chapter 2 Related Work 4 2.1 Chinese Character 4 2.2 DCNN-based classifier and related HCCR methods 5 2.3 Methods applied to Open-set recognition 8 Chapter 3 Methods 10 3.1 Problem definition 10 3.2 Proposed model: pseudo-Siamese network with global classifier 12 3.2.1 Dual-encoder 12 3.2.2 Similarity computation 13 3.2.3 Auxiliary task: global classification 14 3.2.4 Training and Testing 14 Chapter 4 Experiment and Discussion 15 4.1 Datasets 15 4.1.1 Train-HW: Draw from CASIA-HWDB 17 4.1.2 Test-HW: Draw from ICDAR-2013 competition database 17 4.1.3 Test-Ancient-1 and Test-Ancient-2: Draw from CASIA-AHCDB 17 4.1.4 Test-ESUN: Draw from ESUN artificial intelligence 2021 summer challenge dataset 18 4.2 Implementation detail 19 4.3 Experiments on hyperparameters 19 4.3.1 The effectiveness of the auxiliary task 20 4.3.2 The influence of encoder output dimension. 22 4.3.3 The influence of the number of templates at the training stage 22 4.3.4 The influence of template style at training stage 22 4.3.5 Overall performance 23 4.4 Performance comparison with HCCR methods under the closed-set condition 25 4.4.1 Error analysis 26 4.5 Performance comparison with radical-based methods on unseen classes. 27 4.6 The pros and cons of transforming a DCNN classifier into PSN-GC architecture. 28 4.7 Oracle inscription recognition 30 4.7.1 Dataset 30 4.7.2 Data preparation and model training 32 4.7.3 Experiments under the closed-set condition 33 4.7.4 Experiments under the open-set condition 33 4.7.5 Oracle data cleaning 35 Chapter 5 Conclusion 37 Bibliography 39 | |
dc.language.iso | en | |
dc.title | 基於元學習的開集中文字元辨識 | zh_TW |
dc.title | Meta Learning for Open-set Handwritten Chinese Character Recognition | en |
dc.type | Thesis | |
dc.date.schoolyear | 110-2 | |
dc.description.degree | 碩士 | |
dc.contributor.oralexamcommittee | 陳中明(Chung-Ming Chen),張恆華(Herng-Hua Chang),王祥安(Hsiang-An Wang) | |
dc.subject.keyword | 深度學習,卷積神經網路,手寫中文辨識,甲骨文辨識, | zh_TW |
dc.subject.keyword | deep learning,convolution neural network,handwritten Chinese character recognition,oracle bone inscription recognition, | en |
dc.relation.page | 42 | |
dc.identifier.doi | 10.6342/NTU202203489 | |
dc.rights.note | 同意授權(限校園內公開) | |
dc.date.accepted | 2022-09-27 | |
dc.contributor.author-college | 工學院 | zh_TW |
dc.contributor.author-dept | 工程科學及海洋工程學研究所 | zh_TW |
dc.date.embargo-lift | 2022-09-30 | - |
顯示於系所單位: | 工程科學及海洋工程學系 |
文件中的檔案:
檔案 | 大小 | 格式 | |
---|---|---|---|
U0001-1609202219070900.pdf 授權僅限NTU校內IP使用(校園外請利用VPN校外連線服務) | 1.92 MB | Adobe PDF | 檢視/開啟 |
系統中的文件,除了特別指名其著作權條款之外,均受到著作權保護,並且保留所有的權利。