Nordlöf, J.: Comparative Analysis of Models for Real-Time Pattern Recognition (2014) In: 2nd International Conference on Multimedia Analysis and Pattern Recognition (2019) Nguyen, M.-T., Schweyer, A.-V., Le, T.-L., Tran, T.-H., Vu, H.: Preliminary results on ancient cham character recognition from cham inscription images. Kesiman, M., et al.: Benchmarking of document image analysis tasks for palm leaf manuscripts from Southeast Asia. In: Bulletin de l’Ecole française d’Extrême-Orient, vol. Schweyer, A.-V.: Chronologie des inscriptions publiées du Campā. Griffith, A., Lepoutre, A., Southworth, W.A., Thanh, P.: The Inscriptions of Campa at the Museum of Cham Sculpture in Da Nang. Takahashi, K., Takahashi, S., Cui, Y., Hashimoto, M.: Remarks on computational facial expression recognition from HOG features using quaternion multi-layer neural network. In: Intelligent Transportation Systems, vol. Junior, O.L., Delgado, D., Gonçalves, V., Nun, U.: Trainable classifier-fusion schemes: an application to pedestrian detection. In: IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR), vol.
2003 (2003)ĭalal, N., Triggs, B.: Histograms of oriented gradients for human detection.
Simard, P.Y., Steinkraus, D., Platt, J.C.: Best practices for convolutional neural networks applied to visual document analysis. In: Proceedings of the 10th International Conference on Document Analysis and Recognition, pp. Leung, K.C., Leung, C.H.: Recognition of handwritten Chinese characters by combining regularization, fisher’s discriminant and distorted sample generation. Xu, N., Wang, W., Qu, X.: On-line sample generation for in-air written Chinese character recognition based on leap motion controller. Perez, L., Wang, J.: The effectiveness of data augmentation in image classification using deep learning. In: Proceedings of the 4th International Workshop on Historical Document Imaging and Processing, pp. Valy, D., Verleysen, M., Chhun, S., Burie, J.C.: A new Khmer Palm leaf manuscript dataset for document analysis and recognition: SleukRith set. Experiments on both the raw test set and its denoised version show very promising results ( \(64.4\%\) and \(88.5\%\) of F1-score on two test sets respectively). Therefore, we use a simple transfer learning procedure which inherits knowledge from similar or of the same family language. Second, even with the augmented dataset, the fact of training a deep model from scratch could be very long and sometimes cannot meet a good local minimum.
Therefore, some data augmentation techniques will be evaluated and investigated to increase the number and variation of samples in the dataset. Firstly, the current pre-built dataset is still small which is usually a main drawback for deep learning based methods. The aim of this paper is to extend that earlier research to work on noising data. We adapted some automatic recognition methods and conducted experiments on the manually denoised dataset. In our previous work, we have built the first dataset of champ inscription images, manually segmented them in glyphs and annotated by an ancient Cham expert. To conserve Cham heritage as well as to make them widely accessible and readable by users, digitization and recognition of ancient Cham glyphs become necessary. Unfortunately, these inscriptions are being abrasive by the time. Ancient Cham glyphs have mostly appeared in inscriptions on stones at some museums in Vietnam.