Uncurated Image-Text Datasets: Shedding Light on Demographic Bias
Open accessThe DOI matches its Crossref record; 25 of the 25 references it cites were found in a registry or an index.
Abstract
The increasing tendency to collect large and uncurated datasets to train vision-and-language models has raised concerns about fair representations. It is known that even small but manually annotated datasets, such as MSCOCO, are affected by societal bias. This problem, far from being solved, may be getting worse with data crawled from the Internet without much control. In addition, the lack of tools to analyze societal bias in big collections of images makes addressing the problem extremely challenging. Our first contribution is to annotate part of the Google Conceptual Captions dataset, widely used for training vision-and-language models, with four demographic and two contextual attributes. Our second contribution is to conduct a comprehensive analysis of the annotations, focusing on how different demographic groups are represented. Our last contribution lies in evaluating three prevailing vision-and-language tasks: image captioning, text-image CLIP embeddings, and text-to-image generation, showing that societal bias is a persistent problem in all of them. https://github.com/noagarcia/phase
Overview
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Citations
Cites
ImageNet: A large-scale hierarchical image database
Deng, J., Dong, W., Socher, R., Li, L.-J., Li, K., & Fei-Fei, L. (2009) · 2009 IEEE Conference on Computer Vision and Pattern Recognition · doi:10.1109/cvpr.2009.5206848 · Cited by 65,333
High-Resolution Image Synthesis with Latent Diffusion Models
Rombach, R., Blattmann, A., Lorenz, D., Esser, P., & Ommer, B. (2022) · 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr52688.2022.01042 · Cited by 20,302
Denoising Diffusion Probabilistic Models
Ho, J., Jain, A., & Abbeel, P. (2020) · arXiv · doi:10.48550/arxiv.2006.11239 · Cited by 8,712 · cited 1×
“In particular, Stable Diffusion relies on pre-trained CLIP embeddings and Diffusion Models [24] to generate an image in the latent space whose embedding is close to the input prompt embedding.”
MethodVQA: Visual Question Answering
Antol, S., Agrawal, A., Lu, J., Mitchell, M., Batra, D., Zitnick, C. L., & Parikh, D. (2015) · 2015 IEEE International Conference on Computer Vision (ICCV) · doi:10.1109/iccv.2015.279 · Cited by 4,487
The validity and practicality of sun-reactive skin types I through VI
Fitzpatrick, T. B. (1988) · Archives of Dermatology · doi:10.1001/archderm.1988.01670060015008 · Cited by 4,339
ActivityNet: A large-scale video benchmark for human activity understanding
Heilbron, F. C., Escorcia, V., Ghanem, B., & Niebles, J. C. (2015) · 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2015.7298698 · Cited by 2,578
Photorealistic Text-to-Image Diffusion Models with Deep Language Understanding
Saharia, C., Chan, W., Saxena, S., Li, L., Whang, J., Denton, E., Ghasemipour, S. K. S., Ayan, B. K., Mahdavi, S. S., Lopes, R. G., et al. (2022) · arXiv · doi:10.48550/arxiv.2205.11487 · Cited by 2,528 · cited 2×
“This large amount of data has led to training some disruptive models in the field, such as CLIP [37] trained on 400 million image-text pairs; Imagen [41] trained on 860 million image-text pairs; Flamingo [1] trained on 2.3 billion images and short videos paired with text; DALL-E 2 [38] trained on 650 million images; o…”
InfluentialHierarchical Text-Conditional Image Generation with CLIP Latents
Ramesh, A., Dhariwal, P., Nichol, A., Chu, C., & Chen, M. (2022) · arXiv · doi:10.48550/arxiv.2204.06125 · Cited by 2,511 · cited 2×
“Those models have been shown to learn visual and language representations that outperform the previous state-of-the-art on tasks such as zero-shot classification [37] or text-to-image generation [38, 39].”
BackgroundViLBERT: Pretraining Task-Agnostic Visiolinguistic Representations for Vision-and-Language Tasks
Lu, J., Batra, D., Parikh, D., & Lee, S. (2019) · arXiv · doi:10.48550/arxiv.1908.02265 · Cited by 2,109
UNITER: UNiversal Image-TExt Representation Learning
Chen, Y.-C., Li, L., Yu, L., El Kholy, A., Ahmed, F., Gan, Z., Cheng, Y., & Liu, J. (2020) · Lecture Notes in Computer Science · doi:10.1007/978-3-030-58577-8_7 · Cited by 2,055
Cited by
Design Principles for Generative AI Applications
Weisz, J. D., He, J., Muller, M., Hoefer, G., Miles, R., & Geyer, W. (2024) · Proceedings of the CHI Conference on Human Factors in Computing Systems · doi:10.1145/3613904.3642466 · Cited by 232
“…property [47, 107], the circumvention or reverse-engineering of prompts through attacks [35], the production of hateful, toxic, or profane language [60], the disclosure of sensitive or personal information [83], the production of malicious source code [26, 28], and a lack of representation of…”
BackgroundPromptCharm: Text-to-Image Generation through Multi-modal Prompting and Refinement
Wang, Z., Huang, Y., Song, D., Ma, L., & Zhang, T. (2024) · Proceedings of the CHI Conference on Human Factors in Computing Systems · doi:10.1145/3613904.3642803 · Cited by 115
“Moreover, recent studies from the ML community have shown that SOTA generative AI and large language models could actually misinterpret the user’s intention in the input text prompt [20, 32].”
BackgroundPlaygrounds and Prejudices: Exploring Biases in Generative AI For Children.
Baines, A., Gruia, L., Collyer-Hoar, G., & Rubegni, E. (2024) · Proceedings of the 23rd Annual ACM Interaction Design and Children Conference · doi:10.1145/3628516.3659404 · Cited by 21
“The datasets used are, often, large and lack quality control [15, 33], which serves to create further conversation regarding what children are subjected to by allowing some of those tools to be easily accessible by young, vulnerable audiences.”
BackgroundInspecting the Geographical Representativeness of Images from Text-to-Image Models
Basu, A., Babu, R. V., & Pruthi, D. (2023) · 2023 IEEE/CVF International Conference on Computer Vision (ICCV) · doi:10.1109/iccv51070.2023.00474 · Cited by 21
“[16] annotate and study the Conceptual Captions dataset [43] to understand the representation of different demographic groups.”
MethodUnderstanding Gender Bias in AI-Generated Product Descriptions
Kelly, M., Tahaei, M., Smyth, P., & Wilcox, L. (2025) · Proceedings of the 2025 ACM Conference on Fairness, Accountability, and Transparency · doi:10.1145/3715275.3732169 · Cited by 4
Metadata
- Authors
- Garcia, N., Hirota, Y., Wu, Y., & Nakashima, Y.
- Venue
- 2023 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)
- Publisher
- IEEE
- Published
- 2023-06
- Pages
- 6957-6966
- ISSN
- 1063-6919 (ISSN-L)
- OpenAlex
- W4386065345
- Cited by
- 43
- References
- 25
- FWCI
- 11.09
- Type
- conference paper
- Language
- en
- Open access
- green
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
How to cite
Cite CoChat for the checks on this page, and the paper for its own findings.
- This page
- CoChat. (2026). Citation check: Uncurated Image-Text Datasets: Shedding Light on Demographic Bias. https://app.cochat.ai/papers/10.1109/cvpr52729.2023.00672
- The paper
- Garcia, N., Hirota, Y., Wu, Y., & Nakashima, Y. (2023). Uncurated Image-Text Datasets: Shedding Light on Demographic Bias. 2023 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). https://doi.org/10.1109/cvpr52729.2023.00672
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.