# MDW_Net

**Repository Path**: dglxlcl/mdw_net

## Basic Information

- **Project Name**: MDW_Net
- **Description**: An extremely lightwight CNN model which is specially designed for font recognition。
- **Primary Language**: Unknown
- **License**: CC-BY-SA-4.0
- **Default Branch**: master
- **Homepage**: None
- **GVP Project**: No

## Statistics

- **Stars**: 0
- **Forks**: 0
- **Created**: 2025-05-29
- **Last Updated**: 2026-07-22

## Categories & Tags

**Categories**: Uncategorized

**Tags**: None

## README

# 多重深度卷积堆叠网络(MDW_Net）
一种用于字体识别的极致轻量化卷积网络。参数量仅20.7k，比所有主流轻量化卷积模型小得多，在多个数据集上的字体识别任务上取得SOTA效果。

# 引用信息
If you find this repo useful to your project, please consider to cite it with following bib:  
如果你觉得本仓库对于你的研究或项目有用，请考虑使用以下bib进行引用：
```bibtex
@INPROCEEDINGS{11360422,
  author={Liu, Changlong and Zhou, Xuewen and Jiang, Rong and Zhou, Bin and Ma, Dan},
  booktitle={2025 IEEE 10th International Conference on Data Science in Cyberspace (DSC)}, 
  title={MDW-Net: A Lightweight Convolutional Neural Network Model for Chinese Font Recognition}, 
  year={2025},
  volume={},
  number={},
  pages={407-414},
  keywords={Analytical models;Visualization;Image recognition;Accuracy;Convolution;Computational modeling;Computer architecture;Data models;Convolutional neural networks;Character recognition;computer vision;font recognition;Chinese font styles;CNN;structure design;lightweight models},
  doi={10.1109/DSC67331.2025.00059}}
```

## 论文其它章节代码：  
1. 论文第四章所有实验代码见：https://gitee.com/dglxlcl/char_recg  
2. 论文第五章所有实验代码见：https://gitee.com/dglxlcl/cf_ocr.git