Image Aesthetics Assessment Using Graph Attention Network

Ghosal, Koustav; Smolić, Aljoša

doi:10.1109/icpr56361.2022.9956162

Cited by 6 publications

(1 citation statement)

References 33 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…And Dou et al [15] present deep neural network named Webthetics, which is trained from the collected user rating data, can extract representative features from raw webpages and quantify their aesthetics. However, aesthetic assessment methods based on machine learning is less applied to practical products, in this regard, we carry out visual esthetics experiment of automobile headlamps using aesthetic assessment model based on graph attention [16] which is the resent state-of-the-art method to the best of our knowledge.…”

Section: Related Workmentioning

confidence: 99%

Aesthetics Assessment of Automobile Headlights Based on Image Aesthetic Network

Jiang

Liu

Pan

et al. 2023

Preprint

View full text Add to dashboard Cite

Automatically learned aesthetic assessment for images can provide auxiliary value for the fields of art and design. In this study, we explore the aesthetics assessment methods on the visual esthetics of automobile headlamp images, and compare the feasibility of the calculation method of artificial aesthetic degree evaluation. We take the Image Aesthetics Assessment Network using Graph Attention (AAGN) for the automobile headlamps dataset calibration. To enable testing the effectiveness of AAGN, we apply the method of artificial aesthetics degree calculation and entropy weight, as well as assessment methods based on deep Convolutional Neural Network. The results show that the feasibility of the image aesthetic assessment method based on graph attention network is more suitable for users’ subjective aesthetic needs than artificial aesthetics degree calculation.

show abstract

Section: Related Workmentioning

confidence: 99%