Stereotyping and Bias in the Flickr30K Dataset
arXiv:1605.06083
Abstract
An untested assumption behind the crowdsourced descriptions of the images in the Flickr30K dataset (Young et al., 2014) is that they "focus only on the information that can be obtained from the image alone" (Hodosh et al., 2013, p. 859). This paper presents some evidence against this assumption, and provides a list of biases and unwarranted inferences that can be found in the Flickr30K dataset. Finally, it considers methods to find examples of these, and discusses how we should deal with stereotype-driven descriptions in future applications.
In: Proceedings of the Workshop on Multimodal Corpora (MMC-2016), pages 1-4. Editors: Jens Edlund, Dirk Heylen and Patrizia Paggio
Cited by in corpus (6)
- Word Embeddings Quantify 100 Years of Gender and Ethnic Stereotypes
- Women also Snowboard: Overcoming Bias in Captioning Models
- Predicting Demographics, Moral Foundations, and Human Values from Digital Behaviors
- Computer Vision and Conflicting Values: Describing People with Automated Alt Text
- Consolidating Commonsense Knowledge
- On the use of human reference data for evaluating automatic image descriptions