PixTrack: Precise 6DoF Object Pose Tracking using NeRF Templates and Feature-metric Alignment

09/08/2022
by   Prajwal Chidananda, et al.
1

We present PixTrack, a vision based object pose tracking framework using novel view synthesis and deep feature-metric alignment. Our evaluations demonstrate that our method produces highly accurate, robust, and jitter-free 6DoF pose estimates of objects in RGB images without the need of any data annotation or trajectory smoothing. Our method is also computationally efficient making it easy to have multi-object tracking with no alteration to our method and just using CPU multiprocessing.

READ FULL TEXT

page 1

page 2

page 3

research
08/31/2020

Deep Probabilistic Feature-metric Tracking

Dense image alignment from RGB-D images remains a critical issue for rea...
research
06/28/2022

3D Multi-Object Tracking with Differentiable Pose Estimation

We propose a novel approach for joint 3D multi-object tracking and recon...
research
04/21/2020

How to track your dragon: A Multi-Attentional Framework for real-time RGB-D 6-DOF Object Pose Tracking

We present a novel multi-attentional convolutional architecture to tackl...
research
07/24/2019

Efficient Circle-Based Camera Pose Tracking Free of PnP

Camera pose tracking attracts much interest both from academic and indus...
research
11/07/2019

Improving Human Annotation in Single Object Tracking

Human annotation is always considered as ground truth in video object tr...
research
09/29/2022

DirectTracker: 3D Multi-Object Tracking Using Direct Image Alignment and Photometric Bundle Adjustment

Direct methods have shown excellent performance in the applications of v...
research
04/19/2021

Vision-Based Guidance for Tracking Dynamic Objects

In this paper, we present a novel vision-based framework for tracking dy...

Please sign up or login with your details

Forgot password? Click here to reset