This dataset contains visual attribute annotations for over 500 concrete (animate and inanimate) concepts.
All concepts are represented in ImageNet and the feature production norms of McRae et al. (2005). Each concept is annotated with visual attributes based on a taxonomy of 636 attributes. See Silberer et al. (2013) for details.
The download consists of 16 XML files, one per higher-level category (e.g., vehicles, animals).
This dataset contains similarity judgements for 7,576 word pairs representing 500 concrete concepts (the same ones found in ViSA).
All concepts are in ImageNet and the feature production norms of McRae et al. (2005). Each concept occurs in approximately 30 pairs. Similarity ratings were obtained using Amazon Mechanical Turk. Participants were asked to rate a word pair on two dimensions, visual and semantic similarity using a Likert scale of 1 (highly dissimilar) to 5 (highly similar). See Silberer and Lapata (2014) for details.
The download consists of one plain text file listing all concept pairs and mean semantic and visual similarity ratings in one column each.