Handling Object Symmetries in CNN-based Pose Estimation

TL;DR

This paper addresses the challenge of symmetric objects in CNN-based pose estimation by proposing a novel representation called 'closed symmetry loop' that improves handling of symmetries, achieving state-of-the-art results on the T-LESS dataset.

Contribution

The paper introduces the 'closed symmetry loop' representation for better symmetry handling in pose estimation and extends previous algorithms to 6-DOF, improving accuracy on symmetric objects.

Findings

The proposed method effectively handles continuous and discrete symmetries.

It achieves state-of-the-art performance on the T-LESS dataset.

The new representation improves gradient-based optimization for symmetric objects.

Abstract

In this paper, we investigate the problems that Convolutional Neural Networks (CNN)-based pose estimators have with symmetric objects. We considered the value of the CNN's output representation when continuously rotating the object and found that it has to form a closed loop after each step of symmetry. Otherwise, the CNN (which is itself a continuous function) has to replicate an uncontinuous function. On a 1-DOF toy example we show that commonly used representations do not fulfill this demand and analyze the problems caused thereby. In particular, we find that the popular min-over-symmetries approach for creating a symmetry-aware loss tends not to work well with gradient-based optimization, i.e. deep learning. We propose a representation called "closed symmetry loop" (csl) from these insights, where the angle of relevant vectors is multiplied by the symmetry order and then generalize it to 6-DOF. The representation extends our algorithm from [Richter-Klug, ICVS, 2019] including a method to disambiguate symmetric equivalents during the final pose estimation. The algorithm handles continuous rotational symmetry (e.g. a bottle) and discrete rotational symmetry (e.g. a 4-fold symmetric box). It is evaluated on the T-LESS dataset, where it reaches state-of-the-art for unrefining RGB-based methods.