Optimizing Style Recognition Algorithm for Digital Art Images Using Large Language Models (LLMs)
Ye Cao · 2024
In today's era of big data, images have usurped text as the primary content of the Internet, and style analysis of artistic images has emerged as a key area of computer vision research. This study explores the landscape of image style recognition, with a particular focus on digital art images, by harnessing the power of Large Language Models (LLMs). Traditional methods of image style classification, which rely on quantitative standards and image processing techniques, often fall short when faced with complex image content. To address this, personalized classification approaches using unsupervised learning have gained traction, offering flexibility and improved efficiency. In particular, deep learning-based methods have revolutionized fashion style analysis, enabling more accurate recognition and classification through feature extraction. This study proposes a robust style recognition algorithm that utilizes LLMs and ResNet networks to achieve superior performance in digital art image analysis. Experimental results demonstrate the effectiveness of the proposed model, which outperforms traditional approaches such as Generative Adversarial Networks (GANs) and Support Vector Machines (SVMs) in terms of accuracy. This highlights the potential of advanced deep learning techniques in capturing complex image style features.