Abstract: We propose a single-stage, category-level 6-DoF pose estimation algorithm that simultaneously detects and tracks instances of objects within a known category. Our method takes as input the previous and current frame from a monocular RGB video, as well as predictions from the previous frame, to predict the bounding cuboid and 6-DoF pose (up to scale). Internally, a deep network predicts distributions over object keypoints (vertices of the bounding cuboid) in image coordinates, after which a novel probabilistic filtering process integrates across estimates before computing the final pose using PnP. Our framework allows the system to take previous uncertainties into consideration when predicting the current frame, resulting in predictions that are more accurate and stable than single frame methods. Extensive experiments show that our method outperforms existing approaches on the challenging Objectron benchmark of annotated object videos. We also demonstrate the usability of our work in an augmented reality setting.

Paper: arXiv (published at ICRA 2022)

Code: GitHub

Overview:

Pipeline:

We focus on pose tracking (6-DoF translation and rotation) for category-level objects. Given the current RGB image and previous prediction information, the network outputs object location belief in the current frame, then filtering process finds the final pose. The uncertainty estimation is achieved via tracklet-conditioned network and filtering process.

Qualitative Results on the Objectron dataset:

Comparison on a Dynamic Scene:

Augmented Reality Application: