Gem Version Build Status Test Coverage Code Climate Codacy Badge


An unofficial Ruby wrapper for the Clarifai API. This currently uses v2 of the API and was partially created because other Ruby gems are using v1 which is being depricated in late 2017.

Clarifai is an awesome service and hopefully this gem makes it easier to use it and to use AI image tagging in your Ruby apps. Please let me know if there's something you'd like to see but don't or if you see something you wish you didn't. Thanks!

This gem is currently in development and there is lots likely to change so please be aware!


Add this line to your application's Gemfile:

gem 'clarification'

And then execute:

$ bundle

Or install it yourself as:

$ gem install clarification


Configure the gem as you would normally configure a wonderful happy gem.

  Clarification.configure do |config|
    config.api_key = 'a_big_secret_you_got_from_clarifai'
    config.default_public_models = [:food, :general]

This should also work nicely inside a rails initializer.

Each endpoint initialized in your configuration will be called by default from the client. So any predictions using a client configured with :food and :general will make two API requests.

This will change when Workflow is implemented in the gem.


client =
response = client.predict(some_public_url_of_an_image)


The response object returned from the prediction is a hash containing a nicely parsed result for each of the models called. In the above configuration example we have two models. So we will have response[:food] and response[:general] after asking for a prediction. Each of those model results have an object which has three pieces of relevant information: status, concepts, and response_json.

Thusly you can do things like this:

response[:food].concepts.each do |concept|
  if concept.value > 0.90
    puts "#{}"
if response[:food].status.code == 10000
  puts response[:food].status.description

If you want the actual api response, it's here:


All the objects are OpenStructs currently, but I suspect this will change shortly.

As a convenience, the client also maintains the parsed response in the last_response variable.

client.last_response #=> {:food => Objectifier...}

The client has a search object which can be used to index images with the Clarifai application identified by the API key.

Thusly you can do like this:

image_array = [ url_to_picture_of_kitten, url_to_picture_of_pizza, url_to_picture_of_drake]

Once you have your images indexed, you can go and search them by concepts that might be in your pictures as identified by Clarifai's general prediction model.

results ='cat')
results.hits.each do |hit|
  puts hit.url if hit.score > .90

The search client also saves the last search and searched concept as a matter of convenience.

Go and do likewise.


There are three steps to setting up a custom model:

  • Add images with concepts.
  • Create a model with the same concepts that you've associated with images.
  • Train said model.

These three steps can be accomplished thusly:

cat_related_concepts = ["cat", "feline", "superior"]
client.train.add_image(some_url_of_a_cat, cat_related_concepts)
client.train.create_model('cat', cat_related_concepts)

Now that you have the model created and trained, you should be able to predict against it. Except right now, that's not implemented in the Gem... womp womp.


Lots and lots of things. Amongst them:

  • validate public model selections
  • error handling
  • predict multiple images per call
  • predict video
  • better testing
  • documentation
  • use workflows
  • predict against


After checking out the repo, run bin/setup to install dependencies. Then, run rake spec to run the tests. You can also run bin/console for an interactive prompt that will allow you to experiment.

The repo intentionally includes an api key in bin/console to make experimenting easy and quick. If you plan on making a bunch of requests, you can easily sign up for your own free key.


Bug reports and pull requests are welcome on GitHub at This project is intended to be a safe, welcoming space for collaboration, and contributors are expected to adhere to the Contributor Covenant code of conduct.

You can also see various code reporting / CI info here:
travis ci
code climate


The gem is available as open source under the terms of the MIT License.

Code of Conduct

Everyone interacting in the Clarification project’s codebases, issue trackers, chat rooms and mailing lists is expected to follow the code of conduct.