English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Computer-based scene understanding has influenced fields ranging from urbanplanning to autonomous vehicle performance, yet little is known about how wellthese technologies work across social differences. We investigate the biases ofdeep convolutional neural networks (dCNNs) in scene classification, usingnearly one million images from global and US sources, including user-submittedhome photographs and Airbnb listings. We applied statistical models to quantifythe impact of socioeconomic indicators such as family income, Human DevelopmentIndex (HDI), and demographic factors from public data sources (CIA and USCensus) on dCNN performance. Our analyses revealed significant socioeconomicbias, where pretrained dCNNs demonstrated lower classification accuracy, lowerclassification confidence, and a higher tendency to assign labels that could beoffensive when applied to homes (e.g., "ruin", "slum"), especially in imagesfrom homes with lower socioeconomic status (SES). This trend is consistentacross two datasets of international images and within the diverse economic andracial landscapes of the United States. This research contributes tounderstanding biases in computer vision, emphasizing the need for moreinclusive and representative training datasets. By mitigating the bias in thecomputer vision pipelines, we can ensure fairer and more equitable outcomes forapplied computer vision, including home valuation and smart home securitysystems. There is urgency in addressing these biases, which can significantlyimpact critical decisions in urban development and resource allocation. Ourfindings also motivate the development of AI systems that better understand andserve diverse communities, moving towards technology that equitably benefitsall sectors of society.

Digital Divides in Scene Recognition: Uncovering Socioeconomic Biases in Deep Learning Systems