Skip to content

This project is a comprehensive solution for recognizing handwritten digits and text from images, with functionalities for training, testing, and usage, making it suitable for tasks like cheque amount verification and other handwritten text recognition applications.

License

Notifications You must be signed in to change notification settings

theiturhs/handwritten-sentence-prediction

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

5 Commits
 
 
 
 
 
 

Repository files navigation

Handwritten Sentence Prediction

This project is a comprehensive solution for recognizing handwritten digits and text from images, with functionalities for training, testing, and usage, making it suitable for tasks like cheque amount verification and other handwritten text recognition applications.

Row wise and Column wise histogram

This project provides solution for extracting the letters and then predicting the sentence. For this, few assumptions are made:

  1. The handwriting is not cursive.
  2. The text is in english language only.
  3. Written text is human-readable.

Procedure

  1. Training model for digit and character prediction using EMNIST and MNIST dataset available on Kaggle.
  2. Histogram based segmentation to find out the characters and space between characters and words.
  3. Prediction to decode the sentence

Histogram based Segmentation

  1. The original image is converted into binary image.
  2. Column wise sum is calculated to figure out the coordinates where the main contents are present and where space is present.
  3. Two type of spaces are detected: space between subsequent letters and space between words. These are termed as local space and global space respectively. A threshold value is set to distinguish between local and global space.
  4. Row wise sum is calculated to remove any top or bottom space available.
  5. The extracted contents are resized for prediction.
  6. Model prediction is performed.

Example

  1. Binary Image

    Binary Image

  2. Column wise sum - Histogram: frequency vs total number of columns

    Histogram - Column wise

  3. Extracted Characters

    Extracted Characters

  4. Row wise content extraction

    Row wise content extraction

  5. Image resizing

    Resized extracted characters image

  6. Prediction

    Precition

    Predicted Sentence: ONE CAXM FOATY THOOIANO TNENTY SIX

    Actual Text: ONE LAKH FORTY THOUSAND TWENTY SIX

    Note: The accuracy of prediction can be increased by increasing accuracy of trained model.

Application - Cheque amount verification

Decoding amount from extracted text

Decoded amount from predicted sentence

Connect

LinkedIn Gmail Carrd Kaggle GitHub

About

This project is a comprehensive solution for recognizing handwritten digits and text from images, with functionalities for training, testing, and usage, making it suitable for tasks like cheque amount verification and other handwritten text recognition applications.

Topics

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published