Neural networks driven differentiation analysis of art painting works

Fine art paintings are a significant part of cultural heritage, reflecting human creativity and emotional expression through visual representation. The analysis of paintings using artificial intelligence techniques plays a crucial role in both the preservation and innovative development of artistic works. Brushstrokes, as fundamental elements of paintings, carry stylistic and expressive characteristics that are essential for understanding artistic styles. This paper designs an image classification method for paintings that integrates global and local features, where a convolutional neural network is used to obtain an overall stylistic description of an image of a painting. To verify the effectiveness, this paper retrieves painting data from painting databases and the web and constructs a database containing 2040 painting images. The experiments show that the method proposed can effectively analyse the differences of fine art paintings and achieve classification and distinction.