Top AI Repos — open-source AI, indexed and scored
Top AI Repos tracks AI repositories on GitHub and answers two different questions about each one: is it moving right now, and would you bet a product on it.
Top AI Repos tracks AI repositories on GitHub and answers two different questions about each one: is it moving right now, and would you bet a product on it.
✨ 首个CJK（中日韩）字体识别以及样式提取模型 YuzuMarker的字体识别模型与实现 / First-ever CJK (Chinese Japanese Korean) Font Recognition and Style Extractor, side project of YuzuMarker
| Date | Stars |
|---|---|
| 2026-07-24 | 562 |
| 2026-07-25 | 562 |
| 2026-07-28 | 562 |
| 2026-07-30 | 565 |
| 2026-08-06 | 565 |
Today
— stars today
This week
— stars this week
This month
— stars this month
Momentum
0.0
growth rate 0.00%/day
---
title: YuzuMarker.FontDetection
emoji: 😅
colorFrom: blue
colorTo: yellow
sdk: docker
app_port: 7860
---
<div align="center">
<h1>✨YuzuMarker.FontDetection✨</h1>
<p>First-ever CJK (Chinese, Japanese, Korean) font recognition model</p>
<p>
<a href="https://huggingface.co/spaces/gyrojeff/YuzuMarker.FontDetection"><img alt="Click here for Online Demo" src="https://img.shields.io/badge/🤗-Open%20In%20Spaces%20(Online Demo)-blue.svg"/></a>
<img alt="Commit activity" src="https://img.shields.io/github/commit-activity/m/JeffersonQin/YuzuMarker.FontDetection"/>
<img alt="License" src="https://img.shields.io/github/license/JeffersonQin/YuzuMarker.FontDetection"/>
<img alt="Contributors" src="https://img.shields.io/github/contributors/JeffersonQin/YuzuMarker.FontDetection"/>
</p>
<a href="https://www.buymeacoffee.com/gyrojeff" target="_blank"><img src="https://cdn.buymeacoffee.com/buttons/v2/default-yellow.png" alt="Buy Me A Coffee" style="height: 60px !important;width: 217px !important;" ></a>
</div>
## News
- **[Update 2023/05/05] Project recommended on ruanyifeng.com (阮一峰的网络日志 - 科技爱好者周刊): https://www.ruanyifeng.com/blog/2023/05/weekly-issue-253.html**
- **[Update 2023/11/18] Dataset is now opensource! Download on huggingface from here: https://huggingface.co/datasets/gyrojeff/YuzuMarker.FontDetection/tree/master**
## Scene Text Font Dataset Generation
This repository also contains data for automatically generating a dataset of scene text images with different fonts. The dataset is generated using the [CJK font pack by VCB-Studio](https://vcb-s.com/archives/1114) and thousands of background image from [pixiv.net](https://pixiv.net).
The pixiv data will not be shared since it is just randomly scraped. You may prepare your own background dataset that would fit your data distribution as you like.
For the text corpus,
* Chinese is randomly generated from [3500 common Chinese characters](https://gist.github.com/simongfxu/13accd501f6c91e7a423ddc43e674c0f).
* Japanese is randomly generated from a list of lyrics from https://www.uta-net.com.
* Korean is randomly generated from its alphabet.
All text are also mixed with English text to simulate real-world data.
### Data Preparation Walkthrough
1. Download the CJK font pack and extract it to the `dataset/fonts` directory.
2. Prepare the background data and put them in the `dataset/pixivimages` directory.
3. Run following script to clean the file names
```bash
python dataset_filename_preprocess.py
```
### Generation Script Walkthrough
Now the preparation is complete. The following command can be used to generate the dataset:
```bash
python font_ds_generate_script.py 1 1
```
Note that the command is followed by two parameters. The second one is to split the task into multiple partitions, and the first one is the index of the partitioned task to run. For example, if you want to run the task in 4 partitions, you can run the following commands in parallel to speed up the process:
```bash
python font_ds_generate_script.py 1 4
python font_ds_generate_script.py 2 4
python font_ds_generate_script.py 3 4
python font_ds_generate_script.py 4 4
```
The generated dataset will be saved in the `dataset/font_img` directory.
Note that `batch_generate_script_cmd_32.bat` and `batch_generate_script_cmd_64.bat` are batch scripts for Windows that can be used to generate the dataset in parallel with 32 partitions and 64 partitions.
### Final Check
Since the task might be terminated unexpectedly or deliberately by user. The script has a caching mechanism to avoid re-generating the same image.
In this case, the script might not be able to detect corruption in cache (might be caused by terminating when writing to files) during this task, thus we also provides a script checking the generated dataset and remove the corrupted images and labels.
```bash
python font_ds_detect_broken.py
```
After running the script, you might want to rerun the generation script to fill up the hExcerpt of 16,058 characters
Read on GitHub
152
Would you bet a product on this? Bounded 0–100 and slow moving.
matched fp:fa4c863aaec7e217, topic:pytorch
matched fp:fa4c863aaec7e217, topic:computer-vision