Corpus ID: 201871670

Exposing and Correcting the Gender Bias in Image Captioning Datasets and Models

@article{Bhargava2019ExposingAC,
  title={Exposing and Correcting the Gender Bias in Image Captioning Datasets and Models},
  author={Shruti Bhargava and D. Forsyth},
  journal={ArXiv},
  year={2019},
  volume={abs/1912.00578}
}
The task of image captioning implicitly involves gender identification. However, due to the gender bias in data, gender identification by an image captioning model suffers. Also, the gender-activity bias, owing to the word-by-word prediction, influences other words in the caption prediction, resulting in the well-known problem of label bias. In this work, we investigate gender bias in the COCO captioning dataset and show that it engenders not only from the statistical distribution of genders… Expand
6 Citations
Fair Attention-Based Image Captioning
  • PDF
Worst of Both Worlds: Biases Compound in Pre-trained Vision-and-Language Models
  • PDF
Emerging Trends of Multimodal Research in Vision and Language
  • 2
  • PDF

References

SHOWING 1-10 OF 52 REFERENCES
Balanced Datasets Are Not Enough: Estimating and Mitigating Gender Bias in Deep Image Representations
  • 63
  • PDF
Making the V in VQA Matter: Elevating the Role of Image Understanding in Visual Question Answering
  • 777
  • PDF
Age and Gender Estimation of Unfiltered Faces
  • 465
  • PDF
...
1
2
3
4
5
...