论文标题
基于同型相机姿势回归的损失功能
Homography-Based Loss Function for Camera Pose Regression
论文作者
论文摘要
一些最近基于视觉的重新定位算法依靠深度学习方法来从图像数据中进行相机姿势回归。本文着重于嵌入两个姿势之间的误差以执行基于深度学习的相机姿势回归的损失函数。现有的损失功能要么是难以调整的多目标功能，要么是依赖地面真相3D场景点并需要两步训练的不稳定的重新投影错误。为了解决这些问题，我们引入了一种基于多层同构集成的新型损失功能。该新功能不需要事先初始化，仅取决于物理上可解释的超参数。此外，与现有损失功能相比，在良好的重新定位数据集上进行的实验表明，在训练期间，它最小化了均方根再投入误差。
Some recent visual-based relocalization algorithms rely on deep learning methods to perform camera pose regression from image data. This paper focuses on the loss functions that embed the error between two poses to perform deep learning based camera pose regression. Existing loss functions are either difficult-to-tune multi-objective functions or present unstable reprojection errors that rely on ground truth 3D scene points and require a two-step training. To deal with these issues, we introduce a novel loss function which is based on a multiplane homography integration. This new function does not require prior initialization and only depends on physically interpretable hyperparameters. Furthermore, the experiments carried out on well established relocalization datasets show that it minimizes best the mean square reprojection error during training when compared with existing loss functions.
