All posts

Reading image text using Elixir

This post will detail how you can read text off of an image using the Elixir programming language.

To handle this we will use a technology called Optical Character Recognition which is used to find printed or handwritten text characters inside of an image.

The steps we will take to complete this project are:

  1. Install the system package for tesseract (an OCR engine).
  2. Include the tesseract-ocr-elixir lib in your elixir dependencies.
  3. Test the library functionality using IEx.

Important note before installation:

If you don’t install the tesseract engine before trying to use the Elixir library tesseract-ocr-elixir you will likely be met with this error:

iex(1)>".lib/testocr.png") ** (ErlangError) Erlang error: :enoent (elixir 1.12.0) lib/system.ex:1041: System.cmd("tesseract", [".lib/testocr.png", "stdout"], []) (tesseract_ocr 0.1.5) lib/tesseract_ocr.ex:19:

This error is letting you know you haven’t installed tesseract, not that the file path is empty (which without reading the stacktrace you might suspect because of the :enoent eror).

Once you install tesseract this error should go away. You can see the system request to tesseract in the screenshot below:

1. Installing Tesseract

I use Homebrew to install system dependencies on my mac. So for me installing this library was straightforward:

brew install tesseract

The tesseract website has more options for installation.

2. Add tesseract-ocr-elixir lib to deps

In your mix application add the library tesseract-ocr-elixir to deps. This is an Elixir wrapper for OCR.

def deps do [ {:tesseract_ocr, "~> 0.1.5"} ] end

Run mix deps.get to install the package.

3. Test the library functionality

To do a quick test of the library you will need an image with text, so grab your favorite meme and then load your application using iex -S mix.

Now you can test the library by the read function which will print out any words OCR finds:

iex>"test/resources/testocr.png") "OH YOU FOUND SOME INTERNET MEMES YOU MUST BE FUNNY"

That was the last step! You can also use the library to read PDFs and do other fun things.


I built two products to help fund continued efforts for this website, it would mean a lot if you would check them out:
1. 2500+ Words for the Undermotivated Software Engineer - A short essay I wrote on the titled topic ($3).
2. Build 3 Chrome Extensions in Under 2 Hours - A course I built to teach chrome extension development ($9-20).