BCR-Net: A Scale-Adaptive Business Card Recognition Network with Coordinate Attention

View at publisher Summarise with VIRAT
Authors: Xiaojin Chen, Wenshan Xu, Xiaoyang Kuang, Changjun Yu, Fen Yang, Jianrong Yang
Cited by: 0
References: 18

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

Automated Business Card Recognition (BCR) is significant to office automation. However, challenges such as complex layouts, background interference, and significant text scale variance have hindered existing methods from effectively balancing feature extraction and multi-scale fusion. To address these issues, this paper proposes BCR-Net, a robust recognition network leveraging Coordinate Attention (CA) and Adaptive Scale Fusion (ASF). First, considering the high aspect ratios and discrete distribution of business card text, the CA mechanism is incorporated into the backbone. By aggregating bidirectional spatial information, it suppresses background noise while enhancing the model's position awareness and long-range dependency capturing capabilities for elongated text lines. Second, the ASF module is designed to optimize the traditional Feature Pyramid Network (FPN). By utilizing a gating mechanism to dynamically learn fusion weights based on semantic context, it achieves adaptive weighting of deep semantic and shallow detailed features, significantly improving the representation robustness for micro-scale text. Experimental results demonstrate that BCR-Net achieves an F-measure of 88.0 % on a self-collected dataset, a 5.8 % improvement over the baseline. While maintaining high inference efficiency, the proposed method effectively resolves the bottleneck of multi-scale recognition under complex backgrounds, demonstrating superior robustness and practicality.

Fields of Study & Subjects

Handwritten Text Recognition Techniques Vehicle License Plate Recognition Advanced Neural Network Applications

Publication details

Journal / source
International Conference on Electrical Automation and Artificial Intelligence (ICEAAI)
Pages
1207–1211
Publication date
16 January 2026
Language
EN