Emo-AEN: A Lightweight Network for Brand Image Design Based on Aesthetic Evaluation

Abstract

In the aesthetic evaluation of multimedia data, brand image design is closely intertwined with image aesthetics. Although previous researchers have made significant contributions in this field, the intrinsic relationship between the two has not been fully explored. To address this issue, this paper proposes Emo-AEN: a lightweight image aesthetic assessment method that combines brand image design with attention mechanisms. This method takes into account the aesthetic elements involved in the process of brand image design. The network first performs internal fusion operations to obtain fused features of brand image and image aesthetics. Then, through self-attention mechanisms, it thoroughly explores these fused features. This method not only enhances the expressive power of brand image design but also uncovers the intrinsic relationship between brand image design and image aesthetics,and its lightweight model architecture can be deployed in resource-constrained device environments.