Face recognition from a single image per person: A Survey

Face recognition from a single image per person: A Survey

转载请注明:http://blog.csdn.net/stdcoutzyx/article/details/41496773

1 Problem Definition

Given a stored database of faces, the goal is to identify a person from the database later in time in any different and unpredictable poses, lighting, etc. from just one image.

2 Background

Face Recognition techniques can divided into two periods:

  • Geometric-based method period, which is early. For these methods, one sample problem is not a problem.
  • Apperence-based method period, which is now popular. Which is armed with intelligent tools such as computer vision, pattern recognition, machine learning and neural network, making the feature extraction and face recognition
    processes more effective and efficiency.

However, the Apperence-based method suffers the one sample problem.

3 Challenges and Significance of One Sample Problem

Many existing face recognition suffers the one sample problem. These techniques includes:

  • Standard eigenface (PCA).
  • LDA based subspace algorithms.
  • Probabilistic-based eigenface.
  • SVM-based method.
  • Feature line method.
  • Evolution pursuit.
  • Laplacian faces.

To see the paper for detailed explanation.

Significance of the one sample problem:

  • Easy to collect samples.
  • Save storage cost.
  • Save computational cost.

4 Recognizing from one sample per person

  • Holistic method: identify a face using the whole face as input.
  • Local method: use local facial features.
  • Hybrid method: use both local and holistic features to recognize a face.

在下面的内容中,我选择了上表中跟现在研究的算法相关的技术进行说明。

5 Enlarge the training Set

  • Construct new representations

    • Mining more information from the face at hand

  • Create novel visual images
    • Visual sample generation method concentrates on learning extra information from thedomain besides the training set.

6 Local apperence-based method

7 Reported results

8 Reference

[1]. Tan X, Chen S, Zhou Z H, et al. Face recognition from a single image per person: A survey[J]. Pattern Recognition, 2006, 39(9): 1725-1745.

时间: 2024-10-17 02:15:36

Face recognition from a single image per person: A Survey的相关文章

56 Facial Recognition Research Groups to Watch

56 Facial Recognition Research Groups to Watch The research of facial recognition has been a fascinating journey. It began in the 1960s with Woody Bledsoe, Helen Chan Wolf, and Charles Bisson who created programs to assist with basic face recognition

图像处理与计算机视觉基础,经典以及最近发展

*************************************************************************************************************** 在这里,我特别声明:本文章的源作者是   杨晓冬  (个人邮箱:[email protected]).原文的链接是 http://www.iask.sina.com.cn/u/2252291285/ish.版权归 杨晓冬 朋友所有. 我非常感谢原作者辛勤地编写本文章,并愿意共

图像处理与计算机视觉基础,经典以及最近发展--转载

*************************************************************************************************************** 在这里,我特别声明:本文章的源作者是   杨晓冬  (个人邮箱:[email protected]).原文的链接是http://www.iask.sina.com.cn/u/2252291285/ish.版权归 杨晓冬 朋友所有. 我非常感谢原作者辛勤地编写本文章,并愿意共享

transfer learning

作者:刘诗昆链接:https://www.zhihu.com/question/41979241/answer/123545914来源:知乎著作权归作者所有.商业转载请联系作者获得授权,非商业转载请注明出处. Transfer learning 顾名思义就是就是把已学训练好的模型参数迁移到新的模型来帮助新模型训练数据集.就跟其他知友回答的那样,考虑到大部分数据或任务是存在相关性的,所以通过transfer learning我们可以将已经学到的parameter 分享给新模型从而加快并优化模型的学

Local Generic Representation for Face Recognition with Single Sample per Person (ACCV, 2014)

Abstract: 1. 每个类别单个样本的人脸识别(face recognition with single sample per person, SSPP)是一个非常有挑战性的任务,因为在这种情况下很难通过标准样本库(gallery set)里面的样本对待测样本(query sample)的人脸变化量(facial variation)进行预测; 2. 针对这个问题,这篇论文提出了一个基于局部通用表示(local generic representation)框架的人脸识别算法; 3. 该算

Geometric Context from a Single

这个是Derek Hoiem 2005年的文章,也算是对3D信息的利用的一个开启.主页链接是http://www.cs.uiuc.edu/homes/dhoiem/ 翻译呢是为了更好的理解,以后看的时候还可以回头直接看了,还可以随时添加笔记. 摘要 许多CV算法通过忽略图像潜在的3D几何结构而限制了他们的性能.我们展现了通过学习基于外观模型的几何类别来评估场景的粗糙几何属性,甚至是杂乱的自然场景.几何类别描述了图像区域关于相机的3D方向.我们提供了多重假设框架来鲁棒的评估从单张图像获得的场景结构

A Simple Actions Recognition System

1. Problem Definition There's no doubt that researches and applications on the foundation of videos has become a popular field including intelligence surveillance, interactions between human and machines, content-based video retrieval and so on. Howe

CS231n - CNN for Visual Recognition Assignment1 ---- KNN

CS231n - CNN for Visual Recognition Assignment1 -- KNN 这作业怎么这么难,特别是对于我这种刚接触Python的- 反正能做出来的就做,做不出来的我就先抄别人的-.就当加深下对课程理解吧-. k_nearest_neighbor.py中主要有: compute_distances_two_loops compute_distances_one_loop compute_distances_no_loops predict_labels # -*

Very Deep Convolutional Networks for Large-Scale Image Recognition

Very Deep Convolutional Networks for Large-Scale Image Recognition 转载请注明:http://blog.csdn.net/stdcoutzyx/article/details/39736509 这篇论文是今年9月份的论文[1],比較新,当中的观点感觉对卷积神经网络的參数调整大有指导作用,特总结之. 关于卷积神经网络(Convolutional Neural Network, CNN),笔者后会作文阐述之,读者若心急则或可用谷歌百度