Linear-Covariance Loss for End-to-End Learning of 6D Pose Estimation

Most modern image-based 6D object pose estimation methods learn to predict 2D-3D correspondences, from which the pose can be obtained using a PnP solver. Because of the non-differentiable nature of common PnP solvers, these methods are supervised via the individual correspondences. To address this, several methods have designed differentiable PnP strategies, thus imposing supervision on the pose obtained after the PnP step. Here, we argue that this conflicts with the averaging nature of the PnP problem, leading to gradients that may encourage the network to degrade the accuracy of individual correspondences. To address this, we derive a loss function that exploits the ground truth pose before solving the PnP problem. Specifically, we linearize the PnP solver around the ground-truth pose and compute the covariance of the resulting pose distribution. We then define our loss based on the diagonal covariance elements, which entails considering the final pose estimate yet not suffering from the PnP averaging issue. Our experiments show that our loss consistently improves the pose estimation accuracy for both dense and sparse correspondence based methods, achieving state-of-the-art results on both Linemod-Occluded and YCB-Video.

1. Introduction

Estimating the 6D pose of 3D objects from monocular images is a core computer vision task, with many real world applications, such as robotics manipulation [53], [54], autonomous driving [9], [48] and augmented reality [10], [32]. Although this task can be facilitated by the use of RGBD input, depth sensors are not ubiquitous, and thus 6D object pose estimation from RGB images remains an active research area.