A text detection algorithm for natural scenes based on improved GAU and CBAM
Yixin Yu, Yiwei Liang, Yu Wanga, Zenghui Ding, Lixiang Yu, Xingshuo Li · 2023
This paper studies the problems related to text detection in natural scenes, focusing on convolutional neural networks, full convolutional networks, Feature Pyramid Networks (FPN), semantic segmentation, Global Attention Upsample Module (GAU) and Convolutional Block Attention Module (CBAM). Targeting at the low accuracy of current text detection algorithms, a text detection algorithm based on improved GAU and CBAM in natural scene is proposed. The CBAM is first embedded in the GAU module of FPN, and then the improved network is used for text detection base on FPN semantic segmentation. Experimental results show that the network is feasible and stable and the algorithm in this paper has higher accuracy and recall rate compared with similar algorithms.