Label to Region by Bi-Layer Sparsity Priors
来源:互联网 发布:曲柄摇杆机构优化设计 编辑:程序博客网 时间:2024/06/05 11:51
这是我目前找到的最早关于region tagging的文章,这里叫Label to Region。
The basic philosophy is that an image or semantic region can be sparsely reconstructed via the image patches belonging to the images with common image labels。
(1) we cannot directly use visual similarity between the candidate region and atomic patch to select patches for the reconstruction purpose, since an atomic patch is only part of a region and their similarity cannot convey the inclusion relations; and (2) an intuitive way to improve the accuracy of cross-image region-to-region correspondence is to enforce the usage of atomic patches from few images for this reconstruction, from which those selected atomic patches from one image may have high possibility to form a semantic region
The proposed label-to-region assignment process has the following characteristics: 1) the bi-layer sparse coding aims to enforce the usage of merged patches within an image to reconstruct the reference image or semantic region, which ensures the reliability of label propagation; 2) the process does not require exact image object/concept parsing, which is still far from satisfactory on real world images; and 3) no generative model for each label/concept is learnt, and thus it is scalable to applications with large label set.
Two images annotated with common labels are likely to contain some similar patches. However it is generally difficult to directly derive those semantically similar patch pairs between two images
Thus instead we use a group of atomic patches to reconstruct an image or semantic region, and then harness the reconstruction coefficients for propagating the image labels to those localized image patches。(很多patches不能完整地表达一个label ,或者说能同时表达好几个label)
Two images annotated with common labels are likely to contain some similar patches. However it is generally difficult to directly derive those semantically similar patch pairs between two images。(如果不仅仅考虑图片数量少,而且考虑图片的代表性)
- Label to Region by Bi-Layer Sparsity Priors
- Problem with label layer
- Step by Step to Define a BI Publisher Report
- Saliency Detection by Multi-Task Sparsity Puisuit
- Hyperspectral Band Selection by Multitask Sparsity Puisuit
- oracle BI publisher and label
- complementary priors
- label for caffe image data layer
- Copy Objects to Layer
- jump to case label
- CNN: Single-label to Multi-label
- CNN: single-label to multi-label总结
- CNN: Single-label to Multi-label
- BI-Cube to Cube Load
- Why is L1 regularization supposed to lead to sparsity than L2?
- SQL Server BI Step by Step 1
- 论文笔记之:Let there be Color!: Joint End-to-end Learning of Global and Local Image Priors for Automatic
- 论文笔记之:Let there be Color!: Joint End-to-end Learning of Global and Local Image Priors for Automatic
- 二叉树的简单操作
- openstack pep8检查和执行测试用例,及遇到的问题
- RHEL5.6 多网卡bond 绑定
- 进程间通信第三章(System V IPC)
- gentoo xfce桌面fn+f9关闭触摸板
- Label to Region by Bi-Layer Sparsity Priors
- Android 获取ipv4 地址
- mysql不同版本说明
- 食用油做燃料 悉尼大妈将汽车改造成混合动力车
- 据中国政府网消息,国务院总理李克强4月23日主持召开国务院常务会议,确定进一步落实企业投资自主权的政策措施,决定在基础设施等领域推出一批鼓励社会资本参与的项目,部署促进市场公平竞争维护市场正常秩序工作
- OpenGL绘制旋转立方体实例
- 百题纪念之1041 John's trip(欧拉回路)
- activity反复进入onDestroy onCreate方法
- 图像滤波