Open AccessEnglish Letter Recognition Based on TensorFlow Deep Learning
Author(s) -
Lei Wang,
Xin Cao,
Maohua Li,
Jianan Zhao
Publication year - 2020
Publication title -
journal of physics. conference series
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.21
H-Index - 85
eISSN - 1742-6596
pISSN - 1742-6588
DOI - 10.1088/1742-6596/1627/1/012012
Subject(s) - deep learning , artificial intelligence , computer science , transfer of learning , generalization , field (mathematics) , convergence (economics) , process (computing) , machine learning , feature extraction , feature (linguistics) , image (mathematics) , pattern recognition (psychology) , mathematics , mathematical analysis , linguistics , philosophy , pure mathematics , economics , economic growth , operating system
Image recognition has always been a hot research direction. With the continuous progress of technology theory and application, deep learning has a very significant role in the field of image recognition. Today’s image international competitions and enterprise applications are mainly based on deep learning technology, compared with traditional technologies, deep learning is significantly more effective in the application of feature extraction and algorithm models. This paper based on the TensorFlow framework and use deep learning transfer learning fine-tuning to recognize handwritten English characters. In the course of the experiment, the data enhancement method is used to pre-process the collected data, which can increase the amount of training data and improve the generalization ability of the model. At the same time, the parameters and optimizer are continuously optimized to accelerate the convergence speed and finally reach the convergence loss value. Experimental results show that the application of deep learning algorithms has achieved good results in training model feature extraction and recognition accuracy.