Collecting Image Description Datasets using Crowdsourcing
This work provides larger datasets for image description tasks, which is incremental as it builds upon existing datasets like UIUC Pascal Sentence Dataset and Abstract Scenes.
The authors tackled the problem of limited image description data by creating two new datasets with significantly more human descriptions per image than existing ones, using crowdsourcing via Amazon Mechanical Turk.
We describe our two new datasets with images described by humans. Both the datasets were collected using Amazon Mechanical Turk, a crowdsourcing platform. The two datasets contain significantly more descriptions per image than other existing datasets. One is based on a popular image description dataset called the UIUC Pascal Sentence Dataset, whereas the other is based on the Abstract Scenes dataset con- taining images made from clipart objects. In this paper we describe our interfaces, analyze some properties of and show example descriptions from our two datasets.