NeRD: Neural 3D Reflection Symmetry Detector

TL;DR

NeRD introduces a learning-based method for accurately detecting 3D reflection symmetry planes from images, improving over heuristic approaches and aiding tasks like pose estimation and depth regression.

Contribution

The paper presents a novel neural network approach that combines recognition and geometry-based reconstruction to detect 3D symmetry planes more accurately.

Findings

Significantly improved symmetry plane detection accuracy on synthetic and real datasets.

Enhanced downstream task performance such as pose estimation and depth map regression.

Outperforms previous heuristic-based and direct CNN regression methods.

Abstract

Recent advances have shown that symmetry, a structural prior that most objects exhibit, can support a variety of single-view 3D understanding tasks. However, detecting 3D symmetry from an image remains a challenging task. Previous works either assume that the symmetry is given or detect the symmetry with a heuristic-based method. In this paper, we present NeRD, a Neural 3D Reflection Symmetry Detector, which combines the strength of learning-based recognition and geometry-based reconstruction to accurately recover the normal direction of objects' mirror planes. Specifically, we first enumerate the symmetry planes with a coarse-to-fine strategy and then find the best ones by building 3D cost volumes to examine the intra-image pixel correspondence from the symmetry. Our experiments show that the symmetry planes detected with our method are significantly more accurate than the planes from direct CNN regression on both synthetic and real-world datasets. We also demonstrate that the detected symmetry can be used to improve the performance of downstream tasks such as pose estimation and depth map regression. The code of this paper has been made public at https://github.com/zhou13/nerd.