Multi-Task Learning with Low Rank Attribute Embedding for Person Re-identification
来源:互联网 发布:java正则表达式全匹配 编辑:程序博客网 时间:2024/05/21 10:16
ICCV 2015
http://www.umiacs.umd.edu/~fyang/mtl-lorae.html
Person Re-identification 方面的文献,本文主要引入了 person attributes correlation 即人的一些属性具有相关性,利用这些相关性来进行人的检索。
Our contributions:
1)针对 Person Re-identification 提出一个 multi-tasklearning framework
2)在系统嵌入 attributes 信息来提高性能
3)我们提出了一个新颖的目标函数,包含 low level features and attributes
4)我们的实验结果很好!
3.2. Low Rank Attribute Embedding
人的一些属性具有一些很高的相关性,而另一些属性则没有相关性,这里我们用一个相关矩阵来表示这种属性之间的相关性。
The transformation matrix should capture correlations between all attributes pairs since an attribute can be affected by multiple pairs of other attributes globally. Moreover, groups of attributes can be independent from each other, suggesting the low rank property of the transformation matrix。
上面这个和一个公式很相似,贝叶斯公式,后验概率与先验概率和似然函数的乘积成正比。这里的相关性矩阵就是一个先验概率。因为先验概率的存在,对于一些错误的属性标记,我们也能做出一定的修正,减少其影响。
3.3. Multi-Task Learning with Low Rank Attribute Embedding
下面就是主要定义目标函数了,一系列数学方面的推导,最后得到下面的公式:
3.4. Optimization
针对优化工作又进行了一系列推导,能优化的公式如下:
3.5. Re-identification Process
假定搜索库里有C个人,则我们一共有C个class-specific weight matrices and transformation matrices,每个人对应一个 和
对于每个搜索的图像,我们得到一个分数向量,最大那个对应匹配的人。
4 Experiments
效果貌似不错啊!
- Multi-Task Learning with Low Rank Attribute Embedding for Person Re-identification
- A Multi-task Deep Network for Person Re-identification
- Deep Metric Learning for Person Re-Identification
- 行人检索 - Embedding Deep Metric for Person Re-identification
- 阅读A Discriminatively Learned CNN Embedding for Person Re-identification
- 【Person Re-id】A Discriminatively Learned CNN Embedding for Person Re-identification
- 论文研读--Learning Deep Feature Representations with Domain Guided Dropout for Person Re-identification
- 论文阅读:Multi-Scale Triplet CNN for Person Re-Identification
- Learning Mid-level Filters for Person Re-identification
- 【行人识别】Deep Transfer Learning for Person Re-identification
- Deep Ranking for Person Re-identification via Joint Representation Learning
- 【行人识别】Deep Transfer Learning for Person Re-identification
- 【行人识别】Deep Transfer Learning for Person Re-identification
- 【Person Re-ID】Margin Sample Mining Loss: A Deep Learning Based Method for Person Re-identification
- [Paper note] Embedding Deep Metric for Person Re-identification: A Study Against Large Variation
- Embedding Deep Metric for Person Re-identification: A Study Against Large Variations
- 【2016_ECCV】Embedding Deep Metric for Person Re-identification: A Study Against Large Variations
- paper reading——《Improving Person Re-identification by Attribute and Identity Learning》
- 试着开始自己写点东西
- 【SSH项目实战】国税协同平台-16.登录过滤器
- 学习日记:ajax的产生的背景、原理、特点与应用
- Leptonica在VS2010中的编译一,编译它依赖于库
- LeetCode:Plus One
- Multi-Task Learning with Low Rank Attribute Embedding for Person Re-identification
- JavaWeb 输出随机验证码图片
- javascript中判断radio是否选中的方法
- 数字图像处理四大函数库优缺点
- 第十周练习 1010 又一版 A+B
- 关于angular中指令调用controller的方法:&
- Vim设置颜色主题(colorscheme)
- UITextFiled时时获取当前输入的字符长度
- SUN SVM 卷管理