https://github.com/j03-dev/tarzan

Python package for optical character recognition, CNN
https://github.com/j03-dev/tarzan

cnn keras ocr opencv tensorflow

Last synced: 16 days ago
JSON representation

Python package for optical character recognition, CNN

• Host: GitHub
• URL: https://github.com/j03-dev/tarzan
• Owner: j03-dev
• License: mit
• Created: 2022-10-18T11:26:30.000Z (over 2 years ago)
• Default Branch: main
• Last Pushed: 2023-09-29T17:45:32.000Z (over 1 year ago)
• Last Synced: 2024-11-21T23:34:34.759Z (3 months ago)
• Topics: cnn, keras, ocr, opencv, tensorflow
• Language: Python
• Homepage:
• Size: 14.6 KB
• Stars: 0
• Watchers: 2
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

        
# Tarzan

Tarzan no me is a simple package to do optical character recognition with python.

This package is made with python with [tensorflow](https://www.tensorflow.org) develop by google ,

and [opencv-python](https://docs.opencv.org).

this package uses CNN (convolution neuron network) to do image recognition

## Build this package

```bash

python -m pip install --upgrade build

python -m build

```

## Install package

```bash

python -m pip install dist/tarzan-0.0.1-py3-none-any.whl

```

## Train your own Ocr model with dataset

* #### Example

```python

from tarzan import OcrModel

ocr_model = OcrModel(

    'dataset-a-z/data/training_data',

    'dataset-a-z/data/testing_data'

)

ocr_model.train_and_save(path="model_ocr_v1.model")

ocr_model.save_classes(path="classes")

```

### this is the dataset [ocr_dataset](https://www.kaggle.com/datasets/preatcher/standard-ocr-dataset)