Skip to content
This repository has been archived by the owner on Mar 22, 2023. It is now read-only.

Latest commit

 

History

History
74 lines (61 loc) · 3.29 KB

README.md

File metadata and controls

74 lines (61 loc) · 3.29 KB
Raw

Collections-OCR

This repo is archived as of March 2023. The project is not active and the GoogleCloud API + cloudyr R-wrapper has since changed. For more information contact Todd Widhelm

A few scripts that batches of collections label-images through OCR

Google Cloud Vision API & ocrCloudVision.R

Dependencies

Make sure to install these libraries first:

  • googleCloudVisionR - to do OCR magic
  • Other dependencies include readr, tidyr, stringr for data handling

How to run ocrCloudVision.R:

Notes:

  • This currently uses Google's Cloud Vision API, which requires:
  • This can takes over 30 seconds per label-image.
    • Be mindful how many images you add to your "images" directory.
    • Be mindful of your internet connection speed
    • Keep image sizes under 20MB (Overall, smaller image files transfer and process more quickly)
  • Output likely needs some [or many] follow-up/clean-up steps.
    • Batch similar images together to streamline follow-up steps.

To run the script:

  1. Add a folder named "images" to this script's directory
  2. Add the images (JPG & JPEG) you'd like to OCR to that directory
  3. Run the script (Rscript ocrCloudVision.R)

Output from ocrCloudVision.R:

A CSV named "ocrText-[Date-time].csv", containing these columns:

  • "image" = filename for each JPG and JPEG
  • "imagesize" = filesize for each image (in MB)
  • "ocr_start" = start-date and time when an image was submitted to the Google Vision API
  • "ocr_duration" = duration (in seconds) of the OCR process
  • "line_count" = number of lines in each OCR transcription
  • "Line1" - "Line[N]" = text for each line in the OCR transcription of an image.
    • the number of "Line" columns will match the maximum number of lines as needed.

Tesseract & ocrMangle.R

Dependencies

Make sure to install these libraries first:

  • magick - to read in image files
  • tesseract - to do OCR magic
  • stringr - to split the OCR'ed lines to columns

How to run ocrMangle.R:

Notes:

  • This can takes over 10 seconds per label-image.
    • Be mindful how many images you add to your "images" directory.
  • This currently uses Tesseract's English ("eng"), German ("deu"), and Latin ("lat") libraries.
  • Output likely needs some [or many] follow-up/clean-up steps.
    • Batch similar images together to streamline follow-up steps.

To run the script:

  1. Add a folder named "images" to this script's directory
  2. Add the images (JPG & JPEG) you'd like to OCR to that directory
  3. Run the script (Rscript ocrMangle.R)

Output from ocrMangle.R:

A CSV named "ocrText-[Date-time].csv", containing these columns:

  • "image" = filename for each JPG and JPEG
  • "line_count" = number of lines in each OCR transcription
  • "Line1" - "Line[N]" = text for each line in the OCR transcription.
    • the number of "Line" columns will match the maximum number of lines as needed.

Google Drive API & ocrGoogleDrive.R

This is drafty; might work for small batches, but needs work.