TTA-COPE: Test-Time Adaptation for Category-Level Object Pose Estimation

Test-time adaptation methods have been gaining attention recently as a practical solution for addressing source-to-target domain gaps by gradually updating the model without requiring labels on the target data. In this paper, we propose a method of test-time adaptation for category-level object pose estimation called TTA-COPE. We design a pose ensemble approach with a self-training loss using pose-aware confidence. Unlike previous unsupervised domain adaptation methods for category-level object pose estimation, our approach processes the test data in a sequential, online manner, and it does not require access to the source domain at runtime. Extensive experimental results demonstrate that the proposed pose ensemble and the self-training loss improve category-level object pose performance during test time under both semi-supervised and unsupervised settings.

1. Introduction

Object pose estimation is a crucial problem in computer vision and robotics. Advanced methods that focus on diverse variations of object 6D pose estimation have been introduced, such as known 3D objects (instance-level) [28], [38], category-level [18], [36], [43], few-shot [52], and zero-shot pose estimation [13], [47]. These techniques are useful for downstream applications requiring an online operation, such as robotic manipulation [6], [25], [48] and augmented reality [23], [24], [32]. Our paper focuses on the category-level object pose estimation problem since it is more broadly applicable than the instance-level problem.