RNNPose: Recurrent 6-DoF Object Pose Refinement with Robust Correspondence Field Estimation and Pose Optimization

Direct estimating the 6-DoF object pose from a single color image is challenging, and post-refinement is generally needed to achieve high-precision estimation. In this paper, we propose a framework based on a recurrent neural network (RNN) for object pose refinement, which is robust to erroneous initial poses and occlusions. During the recurrent iterations, object pose refinement is formulated as a non-linear least squares problem based on the estimated correspondence field (between a rendered image and the observed image). The problem is then solved by a differentiable Levenberg-Marquardt (LM) algorithm for end-toend training. The correspondence field estimation and pose refinement are conducted alternatively in each iteration to recover accurate object poses. Furthermore, to improve the robustness to occlusions, we introduce a consistencycheck mechanism based on the learned descriptors of the 3D model and observed 2D image, which downweights the unreliable correspondences during pose optimization. Extensive experiments on LINEMOD, Occlusion-LINEMOD, and YCB-Video datasets validate the effectiveness of our method and demonstrate state-of-the-art performance.