Image Feature Extraction and Interactive Design of Cultural and Creative Products Based on Deep Learning

Informatization has brought value recognition in terms of aesthetic feeling of product design and emotional resonance. Interactive Wenchuang (Cultural and Creative) products combine modern interactive technology with traditional Wenchuang forms, which has attracted wide attention. Based on the design concept of interactive Wenchuang products, this article combines deep learning (DL) model and computer aided design (CAD) to extract image features of Wenchuang products. In order to solve the problem of great similarity difference among samples in the stage of feature detection, a similarity preserving method based on depth metric learning is proposed to optimize sample mining and loss. The simulation test shows that the accuracy of image feature detection method based on DL is higher than that of image feature detection method under the condition of normalized image gray level and uncertain angle. The visual inspection system designed in this article can identify the characteristics of products quickly and accurately, and has certain practicability. © 2024 U-turn Press LLC.