The project tries to create software for recognition of a handwritten text from photos (also for Czech language). It uses computer vision and machine learning. And it experiments with different approaches to the problem. It started as a school project which I got a chance to present on Intel ISEF 2018.
Proces of recognition is divided into 4 steps. The initial input is a photo of page with text.
- Detection of page and removal of background
- Detection and separation of words
- Normalization of words
- Separation and recognition of characters (recognition of words)
Main files combining all the steps are OCR.ipynb or OCR-Evaluator.ipynb. Naming of files goes by step representing - name of machine learning model.
git clone https://github.com/Breta01/handwriting-ocr.git
After downloading the repo, you have to download the datasets and models (for more info look into data and models folders).
The project is created using Python 3.6 with Jupyter Notebook. I recommend using Anaconda. If you have it, you can run the installation as:
conda create --name ocr-env --file environment.yml
conda activate ocr-env
Main libraries (all required libraries are in environment.yml):
- Numpy (1.13)
- Tensorflow (1.4)
- OpenCV (3.1)
- Pandas (0.21)
- Matplotlib (2.1)
With all required libraries installed and cloned repo, run jupyter notebook in the directory of the project. Then you can work on the particular notebook.
Best way how to get involved is through creating GitHub issues or solving one! If there aren't any issues you can contact me directly on email.
MIT
If this project helped you or you want to support quick answers to questions and issues. Or you just think it is an interesting project. Please consider a small donation.
