Analysis of the regional styles of artistic patterns based on convolutional neural network utilizing spatial attention mechanism
Abstract
Image recognition has received significant attention in recent years and is bringing about unprecedented changes in the art field. With the rapid development of deep learning and computer vision technology, artificial intelligence can effectively analyze and identify different art styles, thereby providing strong support for multiple areas such as art preservation, creation, and research. Traditional pattern regional style identification relies on the accumulated experience of researchers to establish discrimination standards and determine the regional style of patterns. Due to the complexity and diversity of pattern regional styles, manual identification is usually inefficient and prone to errors. To address these issues, we propose a new method that combines the spatial attention mechanism and a convolutional neural network (CNN) to create a model for identifying the regional styles of patterns, achieving rapid and accurate recognition results. We utilized resources from the Metropolitan Museum of Art, the Smithsonian Institution, and the National Palace Museum (Taipei) to build a self-made dataset, classifying patterns by continent and training the model for recognition. The model has a short recognition time (4.63x10-4 seconds per image) and a high accuracy rate (87%). The above experimental results demonstrate that this method has excellent performance in pattern regional style recognition.