Skip to content

dpm24800/OpenVisionOCR

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

2 Commits
 
 
 
 
 
 
 
 
 
 

Repository files navigation

OpenVisionOCR

OpenVisionOCR is a lightweight Streamlit web app for image-based text extraction with region selection, multi-language OCR (English, Nepali, Hindi), instant translation, one-click copy, and downloadable scan history.

It works like a simple “Google Lens style” crop-and-scan tool focused on fast local usage.


Features

  • Upload images (jpg, jpeg, png)

  • Interactive region selection (crop any part of the image)

  • Full image scan mode

  • OCR for:

    • English
    • Nepali
    • Hindi
  • Optional translation to:

    • English
    • Nepali
    • Hindi
  • One-click copy buttons for:

    • scanned text
    • translated text
  • Scan history stored in session

  • Download scan history as CSV

  • Clean UI with custom styling


Demo Flow

  1. Upload an image
  2. Select a region or enable Full Image Scan
  3. Click Process & Scan
  4. Copy extracted text or translated result
  5. Download your scan history if needed

Tech Stack

  • Streamlit
  • EasyOCR
  • Pillow (PIL)
  • NumPy
  • streamlit-cropper
  • deep-translator
  • Pandas

Installation

Create and activate a virtual environment first.

pip install streamlit easyocr pillow numpy streamlit-cropper deep-translator pandas

Run the app

streamlit run app.py

Supported Languages

OCR

  • English
  • Nepali
  • Hindi

Translation

  • English
  • Nepali
  • Hindi

Auto language detection is used for translation.


How it works

  • The uploaded image is optionally cropped using an interactive cropper.
  • The selected region is converted to a NumPy array.
  • EasyOCR extracts text from the image.
  • The extracted text can be translated using Google Translator.
  • Each scan is stored in Streamlit session state and shown in the history panel.

Output & History

  • The last scans are shown in the Recent Scan History section.
  • History can be exported as a CSV file.
  • Data is stored only for the current session (not persisted to disk).

Limitations

  • History is not saved permanently.
  • OCR accuracy depends heavily on image quality.
  • Large images may be slower to process.

Suggested project structure

openvisionocr/
│
├─ app.py
├─ requirements.txt
└─ README.md

Example requirements.txt

streamlit
easyocr
pillow
numpy
streamlit-cropper
deep-translator
pandas

License

Choose and add a license file (MIT is recommended for open source projects).

About

OpenVisionOCR is a lens style computer vision OCR application built with PyTorch and Streamlit that detects text regions in images, crops them, applies multi language optical character recognition, and visualizes bounding boxes and extracted text through a clean interactive web interface for real world use.

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors

Languages