Captcha recognition

AI & Machine Learning, Python — incorrectly specified categories?

12 USD

Project translated automatically. Log in or register, to view the original

It is necessary to recognize a simple captcha.

The captcha always:

- consists of 5 characters

- characters: uppercase letters of the English alphabet or digits

- characters are located in 99% of cases exactly in the area as in the example

Requirements:

- must be recognized locally (without using services like capsolver and others, without using the network)

- correct recognition: at least 90%, better more

- high recognition speed: up to 100ms per image on a Ryzen 7 CPU (without graphics on board)

- the main script, into which the solution will be integrated, runs on Node.js. When developing the solution, it is necessary to consider the possibility of simple use and interaction between the main script and the captcha solver developed by you.

Perhaps it is not worth reinventing the wheel and using a ready-made solution like https://keras.io/examples/vision/captcha_ocr/ and modifying/training it for needs.
It is also worth paying attention to this thread on Git https://github.com/nbswords/ocr-captchas

I can provide examples in quantity if necessary.

Applications9