- Task 2 - End-to-end Seal Title Recognition - Method: Transformer Seal Text Recognition Networks with Synthetic data and Various Data Enhancements
- Method info
- Samples list
- Per sample details
method: Transformer Seal Text Recognition Networks with Synthetic data and Various Data Enhancements2023-03-13
Authors: Jun Yu Xinjian Gao Yaohui Zhang Ye Pang
Affiliation: University of Science and Technology of China;PingAn Technology
Email: gxjhlj@mail.ustc.edu.cn
Description: 1. We build fully Transformer Networks consisting a Vision Transformer encoder and Transformer decoder,which gives a better performance on curved seal text due to Transformer's powerful global self-attention mechanism.
2. We use a large number of data enhancement strategies, such as random rotation, random blur, brightness adjustment, etc., to simulate the rotation, bending and background text interference in the original training set.
3. We have made a large number of synthetic data sets to make up for the lack of training sets to improve the generalization ability of the model.