Metadata-Version: 2.1
Name: ocr-ops
Version: 0.0.0.4.2.7
Summary: OCR-Ops
Home-page: https://github.com/prateekt/ocr-ops
Author: Prateek Tandon
Author-email: prateek1.tandon@gmail.com
Classifier: Programming Language :: Python :: 3
Classifier: License :: OSI Approved :: MIT License
Classifier: Operating System :: OS Independent
Requires-Python: >=3.9.7
Description-Content-Type: text/markdown
License-File: LICENSE
Requires-Dist: algo-ops
Requires-Dist: easyocr
Requires-Dist: torch (==1.11.0)
Requires-Dist: torchvision (==0.12.0)
Requires-Dist: pytesseract
Requires-Dist: numpy
Requires-Dist: natsort
Requires-Dist: pyspellchecker
Requires-Dist: shapely
Requires-Dist: ezplotly
Requires-Dist: black
Requires-Dist: twine
Requires-Dist: nose2

# ocr-ops

OCR-Ops is infrastructure to perform Optimal Character Recognition (OCR) at scale on a large number of images and videos.
Built on top of the algo-ops framework, OCR-Ops is modular and extensible in its data processing operations.

Key Features:

* Supports building an OCRPipeline that can utilize multiple popular OCR annotation methods (e.g. PyTesseract, EasyOCR,
  etc.) and return the results in structured and efficient fashion within a unified framework.
* Enables multi-levels of information of the OCR application (e.g. text-only, bounding boxes, etc.)
* Allows definition of an image pre-processing pipeline (before OCR) and a text-cleaning pipeline (after OCR) of
  detected but noisy text to enable optimal and robust OCR performance.
* Supports several nice presets that are plug-and-play for the above purpose!
