Access to Clarifai API

Get description of images from Clarifai API. For more information, see Clarifai uses a large deep learning cloud to come up with descriptive labels of the things in an image. It also provides how confident it is about each of the labels.

Find out what is in a (moving) image with perhaps the best off-the-shelf solution: Clarifai provides descriptors for images along with how confident it is about each of the descriptors. It is a bit magical. (See also the 2013 ImageNet LSVR Classification Results.)

To get the current released version from CRAN:


To get the current development version from GitHub:

devtools::install_github("soodoku/clarifai", build_vignettes = TRUE)

To learn about how to use clarifai, see using_clarifai (html version) or analyzing instagram accounts of politicians (html version). There are lots of other interesting ways you can use clarifai. For instance, you can use clarifai for inferring gender of the speaker based on the first name.

You can also launch vignettes within R:

# Using clarifai
vignette("using_clarifai", package = "clarifai")
# Analyzing politician instagram accounts
vignette("poligrams", package = "clarifai")

Scripts are released under MIT License.


clarifai 0.4

  • Rationalized internals. GET and POST abstracted out, except for feedback
  • Better documentation.
  • Supports color for remote images, models
  • Return for image tags was changed by clarifai, adjusted code to handle that
  • Supports languages

clarifai 0.3.1

  • Clarifai removed link to metro north. Fixed that by changing to GitHub.

clarifai 0.3.0

  • Support for color, languages, usage
  • More tests

clarifai 0.2.0

  • Bug fix: cases where fewer than 20 tags are returned now handled nicely
  • More unit tests added
  • New vignette

Reference manual

It appears you don't have a PDF plugin for this browser. You can click here to download the reference manual.


0.4.0 by Gaurav Sood, 6 months ago

Report a bug at

Browse source code at

Authors: Gaurav Sood [aut, cre]

Documentation:   PDF Manual  

Task views: Web Technologies and Services

MIT + file LICENSE license

Imports curl, jsonlite

Suggests knitr, rmarkdown, testthat

See at CRAN