Quickly Solve Captchas with the Lightweight ddddocr Python Library

Hello everyone, I’m Xiao Wu.

Previously I shared a method that combined pillow and pytesseract for captcha recognition. It is free and easy to use, but its accuracy is average, and achieving higher precision usually requires the Baidu API, which involves extra configuration and an internet connection.

A new library called ddddocr (pronounced “dadadi OCR”) has recently been shared in a group and proves to be very practical.

Environment requirements : python >= 3.8<br/>Windows/Linux/macOS Install it with: pip install ddddocr Parameter description :

Using a random captcha image from the web, the library can be tested as follows:

import ddddocr

ocr = ddddocr.DdddOcr()
with open('1.png', 'rb') as f:
    img_bytes = f.read()
res = ocr.classification(img_bytes)
print(res)

The result image shows the recognized text:

The library’s advantages are obvious: the code is extremely concise (only five lines) compared with the previous methods, and it requires no extra environment variables. Using the magic command %%time also shows that the recognition speed is very fast.

Further testing with additional captcha images demonstrates that simple captchas are recognized quickly, though some cases (e.g., the sixth image) suffer from case‑sensitivity issues.

In summary, if you need captcha recognition and do not demand extremely high precision, the ddddocr library is a solid, easy‑to‑use choice.