Image Descriptions' Limitations for People with Visual Impairments

Jandrey, Alessandra Helena; Ruiz, Duncan D.; Silveira, Milene Selbach

doi:10.1145/3472301.3484356

Cited by 6 publications

(5 citation statements)

References 19 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Besides, we identified factors inhibiting manual image descriptions' generation, and recommendations to some of the identified issues. The snowballing results will be presented in the Brazilian Symposium on Human Factors in Computing Systems (IHC-2021) conference [47], which is the main event in the Human-Computer Interaction area in Brazil;…”

Section: Methodsmentioning

confidence: 99%

“…We decided to use a snowballing approach to identify the relevant studies in the literature to answer the following Research Question: "What are the current image descriptions' issues for people with visual impairments?". We followed snowballing procedures defined by Wohlin [99], and our snowballing results will be present at the 2021 Brazilian Symposium on Human Factors in Computing Systems (IHC) [47].…”

Section: Snowballingmentioning

confidence: 99%

See 1 more Smart Citation

Imagery contents descriptions for People with visual impairments

Jandrey,

Ruiz,

Silveira

2023

Proceedings of the XXII Brazilian Symposium on Human Factors in Computing Systems

Self Cite

View full text Add to dashboard Cite

Descrições de imagens visam expressar, em palavras, o conteúdo visual e são essenciais para pessoas que não têm visão. Tais sentenças descritivas são geradas manualmente ou por modelos de Inteligência Artificial (IA). Apesar da sua relevância, a emergência de geradores de descrições automáticas não foi motivada por pessoas com deficiência visual. Portanto, elas ainda causam insatisfação em sua audiência. Neste estudo, nós investigamos problemas em descrições de imagens na literatura por meio da técnica de Snowballing, onde encontramos treze problemas, incluindo aqueles relacionados à Ética, tais como a aparência física, gênero e identidade, raça e deficiência. Nós identificamos cinco razões do porquê pessoas videntes não escrevem descrições para os conteúdos visuais, demonstrando a necessidade de campanhas de acessibilidade para conscientizá-las da importância social das descrições de imagens. Além disso, nós realizamos um conjunto de entrevistas com oito participantes com baixa visão. Nós exploramos as características das sentenças descritivas de 25 imagens de ambientes internos e coletamos as expectativas de descrições de imagens dos participantes. Portanto, através dos resultados do Snowballing e das entrevistas, nós propomos um conjunto de Boas Práticas para auxiliar as ferramentas automáticas e as pessoas videntes na escrita de descrições de imagens de mais satisfatórias e de qualidade. Nós esperamos que os nossos resultados ressaltem a relevância social de sentenças descritivas e encorajam a comunidade a prosseguir com pesquisas interdisciplinares que possam potencialmente minimizar os problemas encontrados no nosso estudo.

show abstract

Section: Methodsmentioning

confidence: 99%

Section: Snowballingmentioning

confidence: 99%

Imagery contents descriptions for People with visual impairments

Jandrey,

Ruiz,

Silveira

2023

Proceedings of the XXII Brazilian Symposium on Human Factors in Computing Systems

Self Cite

View full text Add to dashboard Cite

show abstract

“…Blind users, according to research reports, frequently grapple with the challenges posed by inaccessible visual content displayed on technology devices [37,58]. Designers have responded to these challenges by developing technical solutions to enhance ICT accessibility [36,52,91]. However, accessibility problems cannot always be solved by technical solutions.…”

Section: Blind People and Social Media Accessibilitymentioning

confidence: 99%

"I Got Flagged for Supposed Bullying, Even Though It Was in Response to Someone Harassing Me About My Disability.": A Study of Blind TikTokers’ Content Moderation Experiences

Lyu,

Cai,

Callis

et al. 2024

Proceedings of the CHI Conference on Human Factors in Computing Systems

View full text Add to dashboard Cite

The Human-Computer Interaction (HCI) community has consistently focused on the experiences of users moderated by social media platforms. Recently, scholars have noticed that moderation practices could perpetuate biases, resulting in the marginalization of user groups undergoing moderation. However, most studies have primarily addressed marginalization related to issues such as racism or sexism, with little attention given to the experiences of people with disabilities. In this paper, we present a study on the moderation experiences of blind users on TikTok, also known as "BlindToker, " to address this gap. We conducted semi-structured interviews with 20 BlindTokers and used thematic analysis to analyze the data. Two main themes emerged: BlindTokers' situated content moderation experiences and their reactions to content moderation. We reported on the lack of accessibility on TikTok's platform, contributing to the moderation and marginalization of BlindTokers. Additionally, we discovered instances of harassment from trolls that prompted BlindTokers to respond with harsh language, triggering further moderation. We discussed these findings in the context of the literature on moderation, marginalization, and transformative justice, seeking solutions to address such issues.CCS Concepts: • Human-centered computing → Human computer interaction (HCI); Empirical studies in HCI.

show abstract

“…For instance, accessibility educator and researcher Chancey Fleet described Facebook's automatic image descriptions as "famously useless in the Blind community" despite "garner[ing] a ton of glowing reviews from mainstream outlets" (Fleet, 2021;Hanley et al, 2021). This disconnect appears to stem from a more fundamental mismatch between what PWDs describe as their captioning needs, and what the research community -particularly through its automatic, quantitative evaluationsprioritizes (Jandrey et al, 2021). In particular, surveys with PWDs repeatedly surface the contextual nature of captions.…”

Section: Ethics Statementmentioning

confidence: 99%

VisText: A Benchmark for Semantically Rich Chart Captioning

J.¹,

Boggust²,

Satyanarayan³

2023

Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)

View full text Add to dashboard Cite

Captions that describe or explain charts help improve recall and comprehension of the depicted data and provide a more accessible medium for people with visual disabilities. However, current approaches for automatically generating such captions struggle to articulate the perceptual or cognitive features that are the hallmark of charts (e.g., complex trends and patterns). In response, we introduce VisText: a dataset of 12,441 pairs of charts and captions that describe the charts' construction, report key statistics, and identify perceptual and cognitive phenomena. In VisText, a chart is available as three representations: a rasterized image, a backing data table, and a scene graph -a hierarchical representation of a chart's visual elements akin to a web page's Document Object Model (DOM). To evaluate the impact of VisText, we fine-tune state-of-the-art language models on our chart captioning task and apply prefix-tuning to produce captions that vary the semantic content they convey. Our models generate coherent, semantically rich captions and perform on par with state-of-the-art chart captioning models across machine translation and text generation metrics. Through qualitative analysis, we identify six broad categories of errors that our models make that can inform future work.

show abstract

Image Descriptions' Limitations for People with Visual Impairments

Cited by 6 publications

References 19 publications

Imagery contents descriptions for People with visual impairments

Imagery contents descriptions for People with visual impairments

"I Got Flagged for Supposed Bullying, Even Though It Was in Response to Someone Harassing Me About My Disability.": A Study of Blind TikTokers’ Content Moderation Experiences

VisText: A Benchmark for Semantically Rich Chart Captioning

Contact Info

Product

Resources

About