Lightweight MobileNetV2 offline handwritten Chinese character recognition based on attention mechanism

Pingping Shi; Hongqiong Huang

doi:10.1117/12.2659091

23 November 2022 Lightweight MobileNetV2 offline handwritten Chinese character recognition based on attention mechanism

Pingping Shi, Hongqiong Huang

Proceedings Volume 12454, International Symposium on Robotics, Artificial Intelligence, and Information Engineering (RAIIE 2022); 1245425 (2022) https://doi.org/10.1117/12.2659091
Event: International Symposium on Robotics, Artificial Intelligence, and Information Engineering (RAIIE 2022), 2022, Hohhot, China

Abstract

Over recent years, offline handwritten Chinese character recognition which based on convolutional neural network has become a research hotspot, and the accuracy rate is constantly improving. At the same time, there are problems such as large model size, many calculation parameters and model redundancy. In response to this, this paper proposes a lightweight and improved MobileNetV2 offline handwritten Chinese character recognition model based on attention mechanism. By adding an attention mechanism to the pooling layer of CNN, the L2 norm-constrained Softmax classification function is used in the final classification. In this paper, a research experiment is designed to compare the improved model with the traditional method and the classic CNN model. Experiments show that on the ICDAR-2013 competition test set, the accuracy of this method reaches 96.12%, and the model size is 48.34MB, which is improved in both recognition rate and volume compared with previous methods.

Citation Download Citation

Pingping Shi and Hongqiong Huang "Lightweight MobileNetV2 offline handwritten Chinese character recognition based on attention mechanism", Proc. SPIE 12454, International Symposium on Robotics, Artificial Intelligence, and Information Engineering (RAIIE 2022), 1245425 (23 November 2022); https://doi.org/10.1117/12.2659091

ACCESS THE FULL ARTICLE

INSTITUTIONAL
Select your institution to access the SPIE Digital Library.

SELECT YOUR INSTITUTION

PERSONAL
Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.

PERSONAL SIGN IN

No SPIE Account? Create one

PURCHASE THIS CONTENT

SUBSCRIBE TO DIGITAL LIBRARY

50 downloads per 1-year subscription

Members: $195

Non-members: $335 ADD TO CART

25 downloads per 1 - year subscription

Members: $145

Non-members: $250 ADD TO CART

PURCHASE SINGLE ARTICLE

Includes PDF, HTML & Video, when available