Eine Plattform für die Wissenschaft: Bauingenieurwesen, Architektur und Urbanistik
Eine Plattform für die Wissenschaft: Bauingenieurwesen, Architektur und Urbanistik
Beauty is in the eye of the beholder: Moderate ability to predict perceptions of beauty, restorativeness, and naturalness, in imagery among 10,727 participants
https://orcid.org/0000-0002-8848-1798
https://orcid.org/0000-0003-2213-7959
Highlights We surveyed > 10,000 people to rate image beauty, restorativeness and naturalness. We saw a strong, positive correlation between beauty and restorativeness. It was to accurately predict image perceptions, regardless of demographics. Future research could examine differences by geography or past experiences of users.
Abstract Visual exposure to nature may bolster mental health, possibly through perceptions of scenes as reflecting the concepts of beauty, restorativeness, and naturalness. To date, only a few studies have used image analytics and deep learning methods to attempt to estimate these concepts based on a training set of user rated images. We aimed to 1) test whether the use of deep learning could be applied to predict perceptions of beauty, restorativeness, and naturalness in Google Street View (GSV) imagery (n = 3598 from across the USA); and 2) understand whether researcher-assigned theoretical constructs from the literature predict these concepts in intentionally-selected imagery (n = 6). A national sample of participants (n = 10,727) ranked the qualities of 11 randomly assigned images. We found it difficult to accurately predict ratings of beauty, restorativeness, and naturalness using GSV images. Consistently important image variables for the better-performing models included the total edge contrast index, the interspersion-juxtaposition index, and vertical vegetation. Important rater metrics include the binary Hispanic variable and the time the rater took to complete the survey. We also found similar correlations between the concepts of beauty, restorativeness, and naturalness by sex, urbanicity, and race/ethnicity whereby beauty and restorativeness had the largest correlation. We also found distinct constructs in the intentionally-selected images. Beauty and restorativeness could be predicted by the presence of water, and urban, maintained scenes. Imagery with greenery of various heights was associated with lower beauty and restorativeness and higher naturalness. Pure nature scenes were associated with both higher restorativeness and naturalness. For all three concepts, crowded views were significantly associated with lower values. Distinctly, expansive views were associated with beauty, and scenes with traditional structures were associated with lower naturalness. Future research may usefully explore geographic differences in ratings and the past experiences of participants, as drivers of image perceptions. Accurate prediction of image perceptions and understanding why certain people view or do not view particular images as beautiful, restorative, or natural may inform efforts to promote mental health through urban design.
Beauty is in the eye of the beholder: Moderate ability to predict perceptions of beauty, restorativeness, and naturalness, in imagery among 10,727 participants
https://orcid.org/0000-0002-8848-1798
https://orcid.org/0000-0003-2213-7959
Highlights We surveyed > 10,000 people to rate image beauty, restorativeness and naturalness. We saw a strong, positive correlation between beauty and restorativeness. It was to accurately predict image perceptions, regardless of demographics. Future research could examine differences by geography or past experiences of users.
Abstract Visual exposure to nature may bolster mental health, possibly through perceptions of scenes as reflecting the concepts of beauty, restorativeness, and naturalness. To date, only a few studies have used image analytics and deep learning methods to attempt to estimate these concepts based on a training set of user rated images. We aimed to 1) test whether the use of deep learning could be applied to predict perceptions of beauty, restorativeness, and naturalness in Google Street View (GSV) imagery (n = 3598 from across the USA); and 2) understand whether researcher-assigned theoretical constructs from the literature predict these concepts in intentionally-selected imagery (n = 6). A national sample of participants (n = 10,727) ranked the qualities of 11 randomly assigned images. We found it difficult to accurately predict ratings of beauty, restorativeness, and naturalness using GSV images. Consistently important image variables for the better-performing models included the total edge contrast index, the interspersion-juxtaposition index, and vertical vegetation. Important rater metrics include the binary Hispanic variable and the time the rater took to complete the survey. We also found similar correlations between the concepts of beauty, restorativeness, and naturalness by sex, urbanicity, and race/ethnicity whereby beauty and restorativeness had the largest correlation. We also found distinct constructs in the intentionally-selected images. Beauty and restorativeness could be predicted by the presence of water, and urban, maintained scenes. Imagery with greenery of various heights was associated with lower beauty and restorativeness and higher naturalness. Pure nature scenes were associated with both higher restorativeness and naturalness. For all three concepts, crowded views were significantly associated with lower values. Distinctly, expansive views were associated with beauty, and scenes with traditional structures were associated with lower naturalness. Future research may usefully explore geographic differences in ratings and the past experiences of participants, as drivers of image perceptions. Accurate prediction of image perceptions and understanding why certain people view or do not view particular images as beautiful, restorative, or natural may inform efforts to promote mental health through urban design.
Beauty is in the eye of the beholder: Moderate ability to predict perceptions of beauty, restorativeness, and naturalness, in imagery among 10,727 participants
https://orcid.org/0000-0002-8848-1798
https://orcid.org/0000-0003-2213-7959
Highlights We surveyed > 10,000 people to rate image beauty, restorativeness and naturalness. We saw a strong, positive correlation between beauty and restorativeness. It was to accurately predict image perceptions, regardless of demographics. Future research could examine differences by geography or past experiences of users.
Abstract Visual exposure to nature may bolster mental health, possibly through perceptions of scenes as reflecting the concepts of beauty, restorativeness, and naturalness. To date, only a few studies have used image analytics and deep learning methods to attempt to estimate these concepts based on a training set of user rated images. We aimed to 1) test whether the use of deep learning could be applied to predict perceptions of beauty, restorativeness, and naturalness in Google Street View (GSV) imagery (n = 3598 from across the USA); and 2) understand whether researcher-assigned theoretical constructs from the literature predict these concepts in intentionally-selected imagery (n = 6). A national sample of participants (n = 10,727) ranked the qualities of 11 randomly assigned images. We found it difficult to accurately predict ratings of beauty, restorativeness, and naturalness using GSV images. Consistently important image variables for the better-performing models included the total edge contrast index, the interspersion-juxtaposition index, and vertical vegetation. Important rater metrics include the binary Hispanic variable and the time the rater took to complete the survey. We also found similar correlations between the concepts of beauty, restorativeness, and naturalness by sex, urbanicity, and race/ethnicity whereby beauty and restorativeness had the largest correlation. We also found distinct constructs in the intentionally-selected images. Beauty and restorativeness could be predicted by the presence of water, and urban, maintained scenes. Imagery with greenery of various heights was associated with lower beauty and restorativeness and higher naturalness. Pure nature scenes were associated with both higher restorativeness and naturalness. For all three concepts, crowded views were significantly associated with lower values. Distinctly, expansive views were associated with beauty, and scenes with traditional structures were associated with lower naturalness. Future research may usefully explore geographic differences in ratings and the past experiences of participants, as drivers of image perceptions. Accurate prediction of image perceptions and understanding why certain people view or do not view particular images as beautiful, restorative, or natural may inform efforts to promote mental health through urban design.
Beauty is in the eye of the beholder: Moderate ability to predict perceptions of beauty, restorativeness, and naturalness, in imagery among 10,727 participants
Pearson, Amber L. (Autor:in) / Lin, Zihan (Autor:in) / Shortridge, Ashton (Autor:in)
21.11.2023
Aufsatz (Zeitschrift)
Elektronische Ressource
Englisch
Constructing Artwork - Beauty is in the eye of the beholder.
Online Contents | 2012
AESTHETICS, PART I: BEAUTY IS IN THE MIND OF THE BEHOLDER
| Online Contents | 1996
Scenic Beauty Perceptions Along the ROS
| Online Contents | 1994