Scholarlink

In this paper, we introduce a novel approach to the task of Visual Question Answering (VQA). The highlight of this approach is the integration of both local and global image features, which are then combined with linguistic features to enhance reasoning capabilities and improve the quality of VQA processing. We illustrate the effectiveness and feasibility of this strategy by conducting thorough evaluations on various combinations of these features. Also, we identify combination models with a high potential for future development. These evaluations also pave the way for future research aimed at developing and enhancing accuracy in VQA tasks. The contributions of this study will support researchers by enabling them to utilize the combined power of pre-trained models to enhance image reasoning capabilities in VQA.

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

Integrating Local Features Into Vision Transformer Architecture for Vietnamese Visual Question Answering

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Dang Huynh Cuu Quan,
  • Do Nguyen Dinh Tuan,
  • Tung Le

摘要

In this paper, we introduce a novel approach to the task of Visual Question Answering (VQA). The highlight of this approach is the integration of both local and global image features, which are then combined with linguistic features to enhance reasoning capabilities and improve the quality of VQA processing. We illustrate the effectiveness and feasibility of this strategy by conducting thorough evaluations on various combinations of these features. Also, we identify combination models with a high potential for future development. These evaluations also pave the way for future research aimed at developing and enhancing accuracy in VQA tasks. The contributions of this study will support researchers by enabling them to utilize the combined power of pre-trained models to enhance image reasoning capabilities in VQA.