Driven by digital technology, the research and analysis of art works may require the use of computer vision technology to achieve new research progress. This article focuses on the use of image segmentation and object detection technology to enhance the details and intelligent analysis of art works. By constructing the fusion model, this article combines the image segmentation module based on U-Net with the target detection module based on YOLOv5 (You Only Look Once), and adopts the strategy based on feature fusion to realize their cooperative work. The experiment uses a data set containing 500 different types of art works, which is divided into training set, verification set and test set according to the proportion after professional annotation. In the detail enhancement experiment, the average of the fusion model is 0.92 on SSIM (Structural Similarity Index) and 35.6dB on PSNR (Peak Signal to Noise Ratio), which is significantly better than the traditional method. In the analysis experiment, the average accuracy (mAP) of the fusion model is 0.85, and the average score of art majors is 8.4, which is better than the single model. The results largely indicate that the fusion model has advantages in detail enhancement and intelligent analysis of artworks. It provides an effective solution for the digital research of artworks.
Deng et al. (Sun,) studied this question.