...
首页> 外文期刊>Sensing and imaging >Human-Centric Image Categorization Based on Poselets
【24h】

Human-Centric Image Categorization Based on Poselets

机译:基于Poselet的以人为中心的图像分类

获取原文
获取原文并翻译 | 示例
           

摘要

In daily life, one kind of images are common, in which many people arernpresent and performing certain activities. We call these images human-centricrnimages. As the number of such images gets increasingly larger, to organize andrnaccess them efficiently becomes urgent. Since the categories of human-centricrnimages are determined by human activities in the images, in this paper, we proposernto classify human-centric images by analyzing poses of all humans in them. Specifically,rnfirst, we introduce the notion of poselets, which represent parts of poses ofrnhumans and a method to detect human based on the poselets. Given a human-centricrnimage, to determine its category, we use the poselets and the human detectionrnmethod to detect all possible poselet activations in it and create a statistical representationrnof the poses of humans in the image. Additionally, we also investigatedrnthe influence of contextual information on the categorization of human-centricrnimages. Finally, for evaluating the human-centric image categorization method, fiverncategories of human-centric images are collected from the internet and used forrnexperiments. Experiment results show that the poselet distribution representationsrnare more suitable for representing human-centric images than the popular bag ofrnvisual words method.
机译:在日常生活中,一种图像很常见,其中有很多人在场并进行某些活动。我们称这些图像以人为中心的图像。随着这类图像的数量越来越大,有效地组织和访问它们变得迫在眉睫。由于以人为中心的图像的类别是由图像中的人类活动决定的,因此,本文提出通过分析所有以人为中心的姿态来对以人为中心的图像进行分类。首先,我们介绍姿势的概念,它代表人类姿势的各个部分,以及一种基于姿势的检测人的方法。给定一个以人为中心的图像,要确定其类别,我们将使用体态和人类检测方法来检测图像中所有可能的体态激活,并创建图像中人的姿态的统计表示。此外,我们还研究了上下文信息对以人为中心的图像分类的影响。最后,为了评估以人为本的图像分类方法,从互联网上收集了五种以人为本的图像,并进行了实验。实验结果表明,与流行的袋装视觉单词方法相比,姿态集分布表示更适合表示以人为中心的图像。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号