CRNN OCR

This repository provides you with a deployable model to solve captchas which contains characters with random alphanumeric character which has been strikethrough vertically and/or horizontally.

The tool to scrap these images from the website is also available on google colab

Check out my LinkedIn post for the description, use case, and other relevant information.

Table of Contents

• Type of solvable captchas:
• How to use
  • By Google Notebook
  • By using this repo
    • Use the model provide directly
    • Train the model and then use it
• Accuracy
  • repos test dataset
  • Notebooks
• CONTRIBUTING

Type of solvable captchas:

How to use

By Google Notebook

Make sure you have a Google account and are signed in to access and run the notebook.

1. Go to Colab Notebook

2. Add your folder path to captchasolve function and get a list of strings with predicted texts.

3. Follow the instructions provided in the notebook to run and interact with the project.

By using this repo

1. Download the zip file of the repo and extract it

Use the model provide directly

1. Add your captcha in userset directory present in data Directory
2. Run the model_using.py present in the scripts directory in your python compiler
3. Wait until the file has finished running, then the alphanumaric string after "For user provided captcha: " will be your output

Example

In output:

For user provided captcha:
1/1 [==============================] - 0s 30ms/step
['2DG53S']

Your ans is 2DG53S

Here the captcha was:

Train the model and then use it

(If you want to see/change the training data, u can do so by change the contents of trainset directory present in data Directory)

1. Run the model_training.py present in the scripts directory in your python compiler
2. Then do the steps provieded for Use the model provide directly

Accuracy

You can test the accuracy by either using the repos test dataset or by Notebooks

Using repos test dataset

Use the steps provieded for Use the model provide directly but this time see the output after "For testdataset"

You can see/change the test data in the testset directory present in data Directory

Using notebooks

1. Test the accuracy of model using the notebooks given above on testdataset.

add it in the first line

CONTRIBUTING

Contributions are always welcome! Whether it's a feature request, bug fix, or a new idea, feel free to submit a pull request or open an issue. Let's enhance this tool together! Be sure to read CONTRIBUTING file to know about more contributions details.