Scholarlink

<p>Rangoli, a vibrant and intricate art form integral to Indian festivals, symbolizes cultural heritage and community spirit. Despite its deep-rooted significance, the preservation and classification of rangoli designs have been underexplored. Our work introduces FestivalNet, a new deep learning architecture created to classify festival-specific rangoli patterns. A comprehensive dataset, curated from diverse Indian regions and festivals such as Onam, Diwali, Janmashtami, Pongal, and Ganesh Chaturthi, forms the basis of this work. FestivalNet utilizes fused Mobile Inverted Residual Bottleneck module and dense custom attention blocks to capture complex patterns and hierarchical features inherent in rangoli designs. The proposed model achieves an accuracy of 81.58%, precision of 81.62%, recall of 81.07%, and an F1 score of 81.15%, outperforming competitive architectures like EfficientNetV2_ B0 and MobileNet variants. By bridging tradition with technology, this research aims to preserve the cultural significance of rangoli while fostering appreciation through modern applications.</p>

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

FestivalNet: a deep learning architecture to classify the Indian heritage art-rangoli

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年12月29日

查看全文
  • Sasithradevi A,
  • Sabari Nathan,
  • S. Mohamed Mansoor Roomi,
  • Prakash P

摘要

Rangoli, a vibrant and intricate art form integral to Indian festivals, symbolizes cultural heritage and community spirit. Despite its deep-rooted significance, the preservation and classification of rangoli designs have been underexplored. Our work introduces FestivalNet, a new deep learning architecture created to classify festival-specific rangoli patterns. A comprehensive dataset, curated from diverse Indian regions and festivals such as Onam, Diwali, Janmashtami, Pongal, and Ganesh Chaturthi, forms the basis of this work. FestivalNet utilizes fused Mobile Inverted Residual Bottleneck module and dense custom attention blocks to capture complex patterns and hierarchical features inherent in rangoli designs. The proposed model achieves an accuracy of 81.58%, precision of 81.62%, recall of 81.07%, and an F1 score of 81.15%, outperforming competitive architectures like EfficientNetV2_ B0 and MobileNet variants. By bridging tradition with technology, this research aims to preserve the cultural significance of rangoli while fostering appreciation through modern applications.