dieptran43 / image_text_reader

The module extracts text from image using the tesseract-OCR engine. Generally, text present in the images are blur or are of uneven sizes. The image is pre-processed for better comprehension by OCR. This module first makes bounding box for text in images and then normalizes it to 300 dpi, suitable for OCR engine to read.

image_text_reader

It's a very basic tool to read images , images formatted like a restaurant-menu.

Tesseract-ocr

This tools need tesseract-ocr engine. Help yourself with this --

• https://github.com/tesseract-ocr/tesseract/wiki

Linux

Tesseract is available directly from many Linux distributions. The package is generally called 'tesseract' or 'tesseract-ocr' - search your distribution's repositories to find it. Thus you can install Tesseract 4.x and it's developer tools on Ubuntu 18.x bionic by simply running:

sudo apt install tesseract-ocr
sudo apt install libtesseract-dev

Refer here for more on installation on all other systems.

macOS

Homebrew

To install Tesseract run this command:

brew install tesseract

OCR reads the text extracted image from the full image. Click here

Command to use:

Dockerized image reading

docker run -it yardstick17/image-text-reader bash -c "PYTHONPATH='.' python3 read_image.py read_text_from_local_image -f images/sample_image.jpg"

Read from url

PYTHONPATH='.' python3 read_image.py read_text_from_image_url -u https://marketplace.canva.com/MACHUlPU93Q/1/0/thumbnail_large/canva-peach-green-leaves-garden-vegetarian-pizza-menu-MACHUlPU93Q.jpg

[2017-07-07 16:20:34,119] INFO : Downloading image from url: https://marketplace.canva.com/MACHUlPU93Q/1/0/thumbnail_large/canva-peach-green-leaves-garden-vegeta
[2017-07-07 16:20:35,997] INFO : Saving file: /var/folders/cz/n3vkz7x91qs06nmm9byxxgz00000gr/T/tmpienrxu2c
[2017-07-07 16:20:35,997] INFO : Processing image for text Extraction
[2017-07-07 16:20:36,308] INFO : Removing noise and smoothening image
[2017-07-07 16:20:36,431] INFO : Reading the text inside the contour plotted

Read from local image

PYTHONPATH='.' python3 read_image.py read_text_from_local_image -f images/sample_image.jpg

[2017-07-07 16:32:38,862] INFO : Processing image for text Extraction
[2017-07-07 16:32:39,232] INFO : Removing noise and smoothening image
[2017-07-07 16:32:39,442] INFO : Reading the text inside the contour plotted

Deploy an api for reading text from image!

PYTHONPATH='.' python3 api/app.py

[2017-07-07 16:49:57,818] INFO :  * Running on http://0.0.0.0:6600/ (Press CTRL+C to quit)
[2017-07-07 16:49:57,820] INFO :  * Restarting with stat
[2017-07-07 16:49:58,712] WARNING :  * Debugger is active!
[2017-07-07 16:49:58,738] INFO :  * Debugger pin code: 316-405-633

Sample api deployed on my tiny server. Please be patient with them.

curl -X POST \
  http://54.254.214.96/read_image_from_file/url \
  -F url=https://africatalentbank.com/wp-content/uploads/2014/10/Menu.jpg

Digital Menu

Original Image