Download

We freely distribute the data under the Apache 2.0 license. You can download different parts of the corpus:

• The entire corpus
• Only the transcriptions and annotations
• Only the spoken descriptions
• Only the eye-tracking data

Images

For convenience, we also provide the rescaled images with gray borders. Please note that the images do not fall under the Apache 2.0 license, but each have different Creative Commons licenses. MS COCO provides further information about these licenses.

• Download the images here.
• Download an example of how to plot the eye-tracking and image data here.

Graphical user interfaces

We developed two graphical user interfaces to annotate and explore the data. See this page for more information, or download them here:

• Download the tools pre-loaded with data here. (Reviewers, please choose this one.)
• Download the tools without any data here.

If you would also like to have the uncorrected automatic transcriptions to try out the annotation tool, download them here.

Other data

• Consent forms for the free viewing task and the image description task
(in Dutch)
• Instructions for the image description task
• Instructions for the free viewing task
• Experiment files for the image description task
• Experiment files for the free viewing task
• See the page detailing the creation of DIDEC for our data clean-up and analysis scripts.