FocalPose++: Focal Length and Object Pose Estimation via Render and Compare

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Cífka, Martin, Ponimatkin, Georgy, Labbé, Yann, Russell, Bryan, Aubry, Mathieu, Petrik, Vladimir, Sivic, Josef
Natura: Preprint
Pubblicazione: 2023
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866913572935696384
author Cífka, Martin
Ponimatkin, Georgy
Labbé, Yann
Russell, Bryan
Aubry, Mathieu
Petrik, Vladimir
Sivic, Josef
author_facet Cífka, Martin
Ponimatkin, Georgy
Labbé, Yann
Russell, Bryan
Aubry, Mathieu
Petrik, Vladimir
Sivic, Josef
contents We introduce FocalPose++, a neural render-and-compare method for jointly estimating the camera-object 6D pose and camera focal length given a single RGB input image depicting a known object. The contributions of this work are threefold. First, we derive a focal length update rule that extends an existing state-of-the-art render-and-compare 6D pose estimator to address the joint estimation task. Second, we investigate several different loss functions for jointly estimating the object pose and focal length. We find that a combination of direct focal length regression with a reprojection loss disentangling the contribution of translation, rotation, and focal length leads to improved results. Third, we explore the effect of different synthetic training data on the performance of our method. Specifically, we investigate different distributions used for sampling object's 6D pose and camera's focal length when rendering the synthetic images, and show that parametric distribution fitted on real training data works the best. We show results on three challenging benchmark datasets that depict known 3D models in uncontrolled settings. We demonstrate that our focal length and 6D pose estimates have lower error than the existing state-of-the-art methods.
format Preprint
id arxiv_https___arxiv_org_abs_2312_02985
institution arXiv
publishDate 2023
record_format arxiv
spellingShingle FocalPose++: Focal Length and Object Pose Estimation via Render and Compare
Cífka, Martin
Ponimatkin, Georgy
Labbé, Yann
Russell, Bryan
Aubry, Mathieu
Petrik, Vladimir
Sivic, Josef
Computer Vision and Pattern Recognition
We introduce FocalPose++, a neural render-and-compare method for jointly estimating the camera-object 6D pose and camera focal length given a single RGB input image depicting a known object. The contributions of this work are threefold. First, we derive a focal length update rule that extends an existing state-of-the-art render-and-compare 6D pose estimator to address the joint estimation task. Second, we investigate several different loss functions for jointly estimating the object pose and focal length. We find that a combination of direct focal length regression with a reprojection loss disentangling the contribution of translation, rotation, and focal length leads to improved results. Third, we explore the effect of different synthetic training data on the performance of our method. Specifically, we investigate different distributions used for sampling object's 6D pose and camera's focal length when rendering the synthetic images, and show that parametric distribution fitted on real training data works the best. We show results on three challenging benchmark datasets that depict known 3D models in uncontrolled settings. We demonstrate that our focal length and 6D pose estimates have lower error than the existing state-of-the-art methods.
title FocalPose++: Focal Length and Object Pose Estimation via Render and Compare
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2312.02985