Scholarlink

We present an enhanced deep learning model that advances the state of the art in both artist and artistic style classification. Architectural refinements and optimized training boost artist classification accuracy to 96.3%, reducing the previous error rate by 43% (from 6.5% to 3.7%). Additionally, our model achieves a new benchmark in style classification with 75.3% accuracy, a 17% error reduction over the previous 71.2%. Comprehensive evaluations on large-scale datasets confirm the model’s robustness across diverse visual characteristics and class granularity levels.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

An Enhanced Dual-Stream Architecturefor State-of-the-ArtArtist and Style Classification

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Doron Nevo,
  • Eli David,
  • Nathan S. Netanyahu

摘要

We present an enhanced deep learning model that advances the state of the art in both artist and artistic style classification. Architectural refinements and optimized training boost artist classification accuracy to 96.3%, reducing the previous error rate by 43% (from 6.5% to 3.7%). Additionally, our model achieves a new benchmark in style classification with 75.3% accuracy, a 17% error reduction over the previous 71.2%. Comprehensive evaluations on large-scale datasets confirm the model’s robustness across diverse visual characteristics and class granularity levels.