首页    期刊浏览 2024年12月06日 星期五
登录注册

文章基本信息

  • 标题:Image Finder Mobile Application Based on Neural Networks
  • 本地全文:下载
  • 作者:Nabil M. Hewahi ; Saira Rashid
  • 期刊名称:Brain. Broad Research in Artificial Intelligence and Neuroscience
  • 印刷版ISSN:2067-3957
  • 出版年度:2017
  • 卷号:8
  • 期号:1
  • 页码:5-16
  • 语种:English
  • 出版社:EduSoft publishing
  • 摘要:Nowadays taking photos via mobile phone has become a very important part of everyone’s life. Almost each and every person who has a smart phone also has thousands of photos in their mobile device. At times it becomes very difficult to find a particular photo from thousands of photos, and it takes time. This research was done to come up with an innovative solution that could solve this problem. The solution will allow the user to find the required photo by simply drawing a sketch on the objects in the required picture, for example a tree or car, etc. Two types of supervised Artificial Neural Networks are used for this purpose; one is trained to identify the handmade sketches and other is trained to identify the images. The proposed approach introduces a mechanism to relate the sketches with the images by matching them after training. The experimentation results for testing the trained neural networks reached 100% for the sketches, and 84% for the images of two objects as a case study.
  • 其他摘要:Nowadays taking photos via mobile phone has become a very important part of everyone’s life. Almost each and every person who has a smart phone also has thousands of photos in their mobile device. At times it becomes very difficult to find a particular photo from thousands of photos, and it takes time. This research was done to come up with an innovative solution that could solve this problem. The solution will allow the user to find the required photo by simply drawing a sketch on the objects in the required picture, for example a tree or car, etc. Two types of supervised Artificial Neural Networks are used for this purpose; one is trained to identify the handmade sketches and other is trained to identify the images. The proposed approach introduces a mechanism to relate the sketches with the images by matching them after training. The experimentation results for testing the trained neural networks reached 100% for the sketches, and 84% for the images of two objects as a case study.
  • 关键词:image recognition; neural networks; sketches
国家哲学社会科学文献中心版权所有