Quality of Images Showing Medication Packaging from Individuals with Vision Impairments: Implications for the Design of Visual Question Answering Applications
Abstract
Mobile phone technology empowers individuals with vision impairment to submit images with questions and promptly receive answers from remote humans through visual question answering applications. This study aimed (a) identify challenges for visually impaired users of VizWiz, a visual question answering (VQA) application, to obtain information about medication packaging and (b) recommend design guidelines for better supporting this population to receive visual assistance. Using our four‐category coding scheme, we analyzed the quality of 265 images (131 with questions; 134 without) of medication packages from VizWiz. We found that 46% of the images were legible, 40% contained clear indicators for what information was sought, 40% had minimum background clutter, and only 5% total contained sufficient information to definitively answer the users' questions. We also conducted a thematic analysis to identify major challenges in having their questions answered. Major challenges we identified include: image legibility, incomplete display of key information, and a lack of clarity on what information is being sought. We recommend designers pay more attention to how the technology can facilitate training of users on best practices and provide additional tools to aid visually impaired users in capturing good quality images that can be used to answer their questions.
Integrity signals
Registry
Cites
Statistics
Claims the paper makes
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
Survey respondents provided insight into how product design and packaging affect caption accuracy, confirming prior work regarding medical product packaging.
“Survey respondents also provided insight into how product design and packaging affect caption accuracy, confirming prior work regarding medical product packaging [33].”
From "It's trained by non-disabled people": Evaluating How Image Quality Affects Product Captioning with Vision-Language Models (2026) · cites this paper as [33]
Only 46% of 265 medication package images from the VizWiz dataset were legible.
“Davis et al. [33] analyzed 265 medication package images from the VizWiz dataset and found that only 46% were legible.”
From "It's trained by non-disabled people": Evaluating How Image Quality Affects Product Captioning with Vision-Language Models (2026) · cites this paper as [33]
Recognizing products with rounded labels is generally challenging, as is identifying products from a large panel of text.
“Recognizing products with rounded labels is generally challenging [33], as is identifying products from a large panel of text.”
From "It's trained by non-disabled people": Evaluating How Image Quality Affects Product Captioning with Vision-Language Models (2026) · cites this paper as [33]
Davis et al. showed how medical packaging presents a challenging task for VLMs and is a case where knowing the correct medicine and dosage is critical.
“Davis et al. showed how medical packaging presents a challenging task for VLMs and is a case where knowing the correct medicine and dosage is critical [33].”
From "It's trained by non-disabled people": Evaluating How Image Quality Affects Product Captioning with Vision-Language Models (2026) · cites this paper as [33]
Cites
Using thematic analysis in psychology
Braun, V., & Clarke, V. (2006) · Qualitative Research in Psychology · doi:10.1191/1478088706qp063oa · Cited by 194,414 · see citing works
VizWiz Grand Challenge: Answering Visual Questions from Blind People
Gurari, D., Li, Q., Stangl, A. J., Guo, A., Lin, C., Grauman, K., Luo, J., & Bigham, J. P. (2018) · 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition · doi:10.1109/cvpr.2018.00380 · Cited by 621 · cited 3× · see citing works
“With the intention of making this data openly available, this dataset was anonymized and filtered to remove any private or potentially private material (Gurari et al., 2018).”
BackgroundMethodInfluentialSelf-citationVizWiz: nearly real-time answers to visual questions
Bigham, J. P., Jayant, C., Ji, H., Little, G., Miller, A., Miller, R. C., Miller, R., Tatarowicz, A., White, B., White, S., et al. (2010) · Proceedings of the 23nd annual ACM symposium on User interface software and technology · doi:10.1145/1866029.1866080 · Cited by 601 · see citing works
Kane, S. K., Jayant, C., Wobbrock, J. O., & Ladner, R. E. (2009) · Proceedings of the 11th international ACM SIGACCESS conference on Computers and accessibility · doi:10.1145/1639642.1639663 · Cited by 383 · cited 2× · see citing works
“Assistive technologies before the advent of the smartphone typically only served a singular function, and included tools like optical character recognition (OCR) readers, light detectors, color identifiers, and GPS navigation aids (Bigham et al., 2010; Kane et al., 2009).”
BackgroundAssessing Image Quality Issues for Real-World Problems
Chiu, T.-Y., Zhao, Y., & Gurari, D. (2020) · 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr42600.2020.00370 · Cited by 63 · see citing works
Self-citation
Flagged first, then by citations
Cited by
Sorted by citations
Garg, K., Tang, X., Heo, J., Morgan, D. R., Gergle, D., Sudderth, E. B., & Piper, A. M. (2026) · Proceedings of the 2026 CHI Conference on Human Factors in Computing Systems · doi:10.1145/3772318.3791309 · Cited by 3 · see citing works
“Survey respondents also provided insight into how product de-sign and packaging affect caption accuracy, confirming prior work regarding medical product packaging [33].”
Influential
- Authors
- Davis, N., Xie, B., & Gurari, D.
- Affiliations
- The University of Texas at Austin (US)
- Corresponding author
- Davis, Nathan
- Venue
- Proceedings of the Association for Information Science and Technology
- Publisher
- Wiley
- Published
- 2020-10
- Volume
- 57
- Issue
- 1
- ISSN
- 2373-9231 (ISSN-L)
- OpenAlex
- W3094234586
- Cited by
- 4
- References
- 5
- Citations by year
- 2026: 1 · 2023: 1 · 2022: 1 · 2021: 1
- FWCI
- 1.35
- Type
- journal article
- Language
- en
- Open access
- closed
- Locations
- Proceedings of the Association for Information Science and Technology (publishedVersion)
- Topics
- Social Sciences › Human Factors and Ergonomics › Digital Accessibility for Disabilities
- Health Sciences › Health Professions › Occupational Therapy › Assistive Technology in Communication and Mobility
- Physical Sciences › Computer Science › Human-Computer Interaction › Usability and User Interface Design
- Keywords
- visual question answering, visual impairment, pharmaceutical packaging, image quality, assistive technology, user-generated images, design guidelines, accessibility, background clutter, user training
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
