An improved CNN model based on multi-dimensional feature fusion for the Kansei recognition of product styling in decision-making management systems
提出一种融合颜色、纹理和轮廓特征的改进CNN模型，用于自动识别产品造型的感性类型，实验证明该方法在准确性和效率上优于现有模型，有助于企业自动化产品分类和客户反馈响应。
As the variety of products increases, enterprises must not only accurately identify and provide the product styles expected by customers but also achieve efficient operations in areas such as inventory management and logistics scheduling. To accomplish this, enterprises must be able to categorize products through automation, thereby engancing the overall efficiency of services, operations, and scheduling. The concept of Kansei in product styling refers to the subjective feelings and impressions associated with product design. Identifying Kansei is crucial for assessing different types of product styles. To achieve the automatic identification of product style types, we propose an improved CNN model based on multidimensional feature fusion for Kansei identification for product styling. The results of experimental cases demonstrate that this method can effectively identifies different Kansei types of product styling. The feature fusion strategy, which combining colour, texture and contour with the DRCA-CNN model, surpasses other feature fusion methods and existing models. The research results validate the feasibility of the proposed method and demonstrate an accurate and efficient approach to product styling recognition. This supports automated product classification within production systems, enhances responsiveness to customer feedback, and offers valuable insights for decision-makers and managers in the industrial sector.