ObjectMatch: Robust Registration using Canonical Object Correspondences (CVPR2023) @MatthiasNiessner
ObjectMatch: Robust Registration using Canonical Object Correspondences (CVPR2023)  @MatthiasNiessner
Uploaded April 2023 | Updated November 2026, 2 hours ago
Project: cangumeli.github.io/ObjectMatch

We present ObjectMatch1, a semantic and object-centric camera pose estimator for RGB-D SLAM pipelines. Modern camera pose estimators rely on direct correspondences of overlapping regions between frames; however, they cannot align camera frames with little or no overlap. In this work, we propose to leverage indirect correspondences obtained via semantic object identification. For instance, when an object is seen from the front in one frame and from the back in another frame, we can provide additional pose constraints through canonical object correspondences. We first propose a neural network to predict such correspondences on a per-pixel level, which we then combine in our energy formulation with state-of-the-art keypoint matching solved with a joint Gauss-Newton optimization. In a pairwise setting, our method improves registration recall of state-of-the-art feature matching, including from 24% to 45% in pairs with 10% or less inter-frame overlap. In registering RGB-D sequences, our method outperforms cutting-edge SLAM baselines in challenging, low-frame-rate scenarios, achieving more than 35% reduction in trajectory error in multiple scenes.
ObjectMatch: Robust Registration using Canonical Object Correspondences (CVPR2023)GaussianAvatars: Photorealistic Head Avatars with Rigged 3D GaussiansVideo Diffusion Avatars: Taming Video Diffusion Models into Controllable 3D AvatarsIntrinsic Image Diffusion for Single-view Material EstimationDeleted videoDeleted videoMonoNPHM: Dynamic Head Reconstruction from Monocular VideosInverse Path Tracing for Joint Material and Lighting Estimation (CVPR 2019 Oral)TUM AI Lecture Series - Neural Fields Beyond Novel View Synthesis (Andrea Tagliasacchi)DiffusionAvatars: Deferred Diffusion for High-fidelity 3D Head AvatarsDeleted videoNeural Non-Rigid Tracking (NeurIPS 2020)
Matthias Niessner |

ObjectMatch: Robust Registration using Canonical Object Correspondences (CVPR'2023)

SHARE TO X SHARE TO REDDIT SHARE TO FACEBOOK WALLPAPER