PoseNet: A Convolutional Network for Real-Time 6-DOF Camera Relocalization

https://zhuanlan.zhihu.com/p/53044278

https://zhuanlan.zhihu.com/p/45483149

abstract: 训练了一个神经网络可以从单张图像端到端的回归出相机的位置和姿态，有别于SLAM，不再需要额外的工程操作或者图优化。室外精度达到2m，3°。室内精度达到0.5m，5°偏差。网络是23层，利用transfer learning from recognition to re-localization 在目标分类的网络上pre-train的模型。比依赖sift关键点提取，匹配的方法更robust。

Contribution:

1）利用transfer learning 任务从目标识别，迁移到re-localization。

利用structure from motion根据图像序列/视频，自动生成训练label（camera pose），减少了人类标注的工作。

3）避免传统SLAM的pipeline: 比如需要存储densely spaced keyframes, appearance-based localization, landmarked-based pose estimation, frame-to-frame feature correspondence.

Loss function:

作者实验发现把位置和姿态分成两个网络进行训练的效果并不好，猜测是位置和姿态的耦合关系，所以还是要放在一起训练。

网络结构：

GoogLeNet pre-trained for classification 改造成regression问题。

replace the three softmax classfiers with affine regressors.
在最终输出层之前插入了一个全连接层，before the regressor.用来当作local feature vector.

一个比较有趣的应用就是你在指定区域拍张照后, 然后得到google map里街景的一个结果
http://mi.eng.cam.ac.uk/projects/relocalisation/#results

PoseNet: A Convolutional Network for Real-Time 6-DOF Camera Relocalization相关推荐

快速多尺度人脸检测2--Multi-Scale Fully Convolutional Network for Face Detection in the Wild
Multi-Scale Fully Convolutional Network for Face Detection in the Wild CVPRW2017 上一篇文献的姊妹篇吧,如何能够快速的实 ...
快速多尺度人脸检测--Multi-Scale Fully Convolutional Network for Fast Face Detection
Multi-Scale Fully Convolutional Network for Fast Face Detection BMVC 2016 如何能够快速的实现多尺度人脸检测了? 本文的思路是 ...
人群密度估计--CrowdNet: A Deep Convolutional Network for Dense Crowd Counting
CrowdNet: A Deep Convolutional Network for Dense Crowd Counting published in the proceedings of ACM ...
语义分割--Large Kernel Matters--Improve Semantic Segmentation by Global Convolutional Network
Large Kernel Matters–Improve Semantic Segmentation by Global Convolutional Network https://arxiv.org ...
论文笔记（SocialGCN: An Efficient Graph Convolutional Network based Model for Social Recommendation）
一个有效的基于图卷积神经网络的社交推荐模型原文链接:SocialGCN: An Efficient Graph Convolutional Network based Model for Socia ...
人脸关键点: DCNN-Deep Convolutional Network Cascade for Facial Point Detection
近期学习关键点检测相关内容,基于CNN的方法已经替代以往经典的方法(ASM,AAM等),于是乎得看看CNN是如何应用到关键点检测上的. 创新点: 1.将CNN应用到人脸关键点检测当中 2.提出级联CN ...
【U-net】【粗读】Retinal blood vessel segmentation using fully convolutional network with transfer...
Retinal blood vessel segmentation using fully convolutional network with transfer learning 摘要介绍数据集 ...
paper reading：Part-based Graph Convolutional Network for Action Recognition
paper reading:Part-based Graph Convolutional Network for Action Recognition 文章目录 paper reading:Part- ...
FCN（Fully Convolutional Network）与Unet：谈到语义分割不得不提的两个网络
转载请注明出处:https://blog.csdn.net/wyzjack47/article/details/81107980 语义分割经典之作:FCN(Fully Convolutional Ne ...

PoseNet: A Convolutional Network for Real-Time 6-DOF Camera Relocalization

PoseNet: A Convolutional Network for Real-Time 6-DOF Camera Relocalization相关推荐

最新文章

热门文章