Image to Text OCR: Chinese, Japanese, Korean
Pricing
from $1.50 / 1,000 image reads
Image to Text OCR: Chinese, Japanese, Korean
Read text from image links, with a confidence score per image. Traditional and Simplified Chinese, Japanese, Korean and English. Pay per image; failed images cost nothing. 圖片文字辨識 OCR：繁體中文、简体中文、日本語、한국어。
Pricing
from $1.50 / 1,000 image reads
Rating
0.0
(0)
Developer
Inkbridge Studio
Maintained by CommunityActor stats
0
Bookmarked
2
Total users
1
Monthly active users
a day ago
Last modified
Categories
Share
Image to Text OCR: Chinese, Japanese, Korean, English
Give it image links. Get the text back, with a confidence score for each image. Built for Traditional and Simplified Chinese, Japanese, Korean and English, so it suits receipts, screenshots, product photos, signs and scanned pages.
中文、日本語、한국어
- 繁體中文 OCR： 給圖片網址，把圖片裡的文字轉成文字，支援繁體中文、简體中文、日文、韓文與英文。每張 US$0.002，讀不到的圖不收費。
- 简体中文 OCR： 提供图片链接，识别图片中的文字，支持简体中文、繁体中文、日文、韩文和英文。每张 US$0.002，读取失败不收费。
- 日本語 OCR： 画像の URL から文字を読み取ります。日本語、中国語、韓国語、英語に対応。1 枚 US$0.002、読み取れなかった画像は無料です。
- 한국어 OCR: 이미지 URL에서 글자를 추출합니다. 한국어, 중국어, 일본어, 영어를 지원합니다. 이미지당 US$0.002, 읽지 못한 이미지는 무료입니다.
Use cases
- Pull text from product photos and labels you have the right to use.
- Read screenshots of Chinese, Japanese or Korean apps and web pages, ready for translation.
- Extract text from menus, posters and signs.
- Prepare image text in Asian languages for search, RAG or LLM pipelines.
What it does
- Reads PNG, JPEG, WebP, TIFF, BMP and GIF images from public URLs (first frame of a GIF).
- Returns plain text and an average confidence from 0 to 100, so you can flag weak results.
- Tells you which images failed and why. One bad link does not stop the others.
How to use it
- Add image links to Image URLs.
- Pick OCR languages. Choose only what your images contain; extra languages run slower and can lower accuracy.
- Run it. You get one row per image.
Pricing
- US$0.002 per image read.
- An image that cannot be read costs nothing.
- Each run also has a US$0.00005 start fee.
Output
| Field | Meaning |
|---|---|
| url | The image link |
| text | The text found |
| confidence | Average confidence, 0 to 100 |
| width, height | Image size in pixels |
| error | Why the image failed, otherwise empty |
Limits
- Links must be public. 25 MB per image.
- Clean printed text works best. Handwriting, stylised logos, heavy glare and very small text read poorly.
- Text only: no layout, tables or image descriptions. For PDFs, use our PDF to Markdown & Text with OCR Actor.
Run it from code
Call it through the Apify API with this input:
Language codes: eng, chi_tra (Traditional Chinese), chi_sim (Simplified Chinese), jpn, kor.
FAQ
Does it translate? No. It returns the text in the original language.
How do I spot bad results? Check confidence. In our tests a clean Chinese product image scored about 81, and a dense lab table scored about 64 with some garbled lines.
Which engine does it use? Tesseract, running inside the Actor. Images are not sent to a third-party OCR service.
Data
Images are processed only to produce your output, which stays in your own Apify storage. Use only images you have the right to process, and be careful with images that show personal information.
Questions or an image that fails? Open an issue on this Actor's Issues tab.
