Paper
16 October 2023 Optimization of ancient Chinese characters recognition in particular conditions
Cheng Lyu, Yuannan Liu, Chen Wu, Yinghao Xu
Author Affiliations +
Proceedings Volume 12803, Fifth International Conference on Artificial Intelligence and Computer Science (AICS 2023); 1280337 (2023) https://doi.org/10.1117/12.3009377
Event: 2023 5th International Conference on Artificial Intelligence and Computer Science (AICS 2023), 2023, Wuhan, China
Abstract
This paper discusses optimization of ancient Chinese characters recognition in particular conditions. It aims to improve the image recognition technique exerted on ancient Chinese text. During the process, we first use images under good conditions to train the model, AlexNet and ResNet, and predict the input images. Assuming the accuracy rate is over 70%, the image is identified as one in good condition. Then, the results of the models are used as output. If the accuracy does not achieve the expected rate, the images will be filtered and inputted into the models trained by the filtered picture in bad condition to predict. After several epochs of training the two models, ResNet is more appropriate for the process discussed above. Due to its high accuracy of 89%, ResNet is chosen as the model being used in the recognition process. Overall, ancient Chinese characters recognition is improved by process mentioned above.
(2023) Published by SPIE. Downloading of the abstract is permitted for personal use only.
Cheng Lyu, Yuannan Liu, Chen Wu, and Yinghao Xu "Optimization of ancient Chinese characters recognition in particular conditions", Proc. SPIE 12803, Fifth International Conference on Artificial Intelligence and Computer Science (AICS 2023), 1280337 (16 October 2023); https://doi.org/10.1117/12.3009377
Advertisement
Advertisement
RIGHTS & PERMISSIONS
Get copyright permission  Get copyright permission on Copyright Marketplace
KEYWORDS
Education and training

Tunable filters

Matrices

Image filtering

Gaussian filters

Data modeling

Image processing

Back to Top