View the latest news and breaking news today for U.S., world, weather, entertainment, politics and health at CNN.com.

2024年2月1日 · We propose a Transformer-Convolution with Meta-Learning Network (TCML-PIAA) for personalized image aesthetics assessment which effectively combines the strengths of CNN, ViT, and meta -learning to alleviate the challenges posed by differences in aesthetic preferences and the insufficient labeled data for PIAA.

2024年10月16日 · 论文介绍了在图像美学评估（Image Aesthetic Assessment, IAA）任务中，Transformers 模型 相对于卷积神经网络（CNN）模型的局限性，并提出了一种新的解决方案——EAT（Enhancer for Aesthetics-Oriented Transformers）。

2022年3月31日 · To solve the dilemma, we conduct so far, the most comprehensive subjective study of personalized image aesthetics and introduce a new Personalized image Aesthetics database with Rich Attributes (PARA), which consists …

2024年6月21日 · Image Aesthetic Assessment (IAA) methods can be classified under two main categories according to the intensity of their subjectivity: Personalized Image Aesthetic Assessment (PIAA) and Generalized Image Aesthetic Assessment (GIAA). GIAA extrapolates most people's average aesthetic view of an image.

4 天之前 · The Pennsylvania Interscholastic Athletic Association, Inc. (PIAA) is a non-profit corporation organized to eliminate abuses, establish uniform rules, and place interscholastic athletics in the overall context of secondary education.

为了应对这些挑战，我们提出了一种创新网络，将 Transformer 和卷积神经网络 (CNN) 的强大功能与 PIAA 元学习 (TCML-PIAA) 相结合。 首先，我们利用 Vision Transformer 块和 CNN 来提取长期和短期依赖关系，从这两个分支中挖掘更丰富和异构的美学属性。

2024年2月1日 · 实验结果表明，TCML-PIAA在基准数据库上优于现有的最先进方法。 这项研究的创新点在于将Transformer和CNN相结合，利用元学习方法进行个性化图像美学评估，从而解决了传统通用模型无法捕捉每个人独特偏好的问题。

In order to learn the shared prior knowledge when different people judge aesthetics, that is, learn how people judge image aesthetics, we propose a PIAA method based on meta-learning with bilevel gradient optimization (BLG-PIAA), which is trained using individual aesthetic data directly and generalizes to unknown users quickly.

2022年10月9日 · 首先利用协同过滤方法设计了一个带有用户与图像关系编码输入的卷积神经网络 (convolutional neural network，CNN)模型来训练GIAA模型，然后提出一种注意力机制，通过融合图像和用户评论信息来捕获用户对图像语义标签和兴趣区域，最后得到符合用户个性化审美偏好 ...