Camera Relocalization by Exploiting Multi-View Constraints for Scene Coordinates Regression 2019论文笔记
研究机构：阿德莱德大学 本文提出了基于场景坐标回归网络的相机重定位方法，在DSAC++的基础上显式的引入了多视角几何约束，其实就是在训练阶段使用图像对作为输入，并重新设计了三个损失函数，和DSAC++的重投影误差结合，加快了训练的收敛速度，且在室内外数据集上达到了优于DSAC++（无3Dmodel）的精度,但是仍然劣于DSAC++（有3Dmodel）。 论文方法 
相关文章
- 1. Backtracking Regression Forests for Accurate Camera Relocalization
- 2. Relative Geometry-Aware Siamese Neural Networkfor 6DOF Camera Relocalization 2019 论文笔记
- 3. PoseNet: A Convolutional Network for Real-Time 6-DOF Camera Relocalization 2015论文笔记
- 4. PoseNet: A Convolutional Network for Real-Time 6-DOF Camera Relocalization论文阅读
- 5. Camera Relocalization by Computing Pairwise Relative Poses Using Convolutional Neural Network，2017论文
- 6. PoseNet2：Geometric loss functions for camera pose regression with deep learning 论文笔记
- 7. Dual Attention Network for Scene Segmentation 论文笔记
- 8. GN-Net: The Gauss-Newton Loss for Multi-Weather Relocalization 论文笔记 2019
- 9. 【论文笔记】Deep Direct Regression for Multi-Oriented Scene Text Detection
- 10. Efficient multi-output scene coordinate prediction for fast and accurate camera...论文笔记