Abstract:3D symmetry detection is a fundamental problem in computer vision and graphics. Most prior works detect symmetry when the object model is fully known, few studies symmetry detection on objects with partial observation, such as single RGB-D images. Recent work addresses the problem of detecting symmetries from incomplete data with a deep neural network by leveraging the dense and accurate symmetry annotations. However, due to the tedious labeling process, full symmetry annotations are not always practically available. In this work, we present a 3D symmetry detection approach to detect symmetry from single-view RGB-D images without using symmetry supervision. The key idea is to train the network in a weakly-supervised learning manner to complete the shape based on the predicted symmetry such that the completed shape be similar to existing plausible shapes. To achieve this, we first propose a discriminative variational autoencoder to learn the shape prior in order to determine whether a 3D shape is plausible or not. Based on the learned shape prior, a symmetry detection network is present to predict symmetries that produce shapes with high shape plausibility when completed based on those symmetries. Moreover, to facilitate end-to-end network training and multiple symmetry detection, we introduce a new symmetry parametrization for the learning-based symmetry estimation of both reflectional and rotational symmetry. The proposed approach, coupled symmetry detection with shape completion, essentially learns the symmetry-aware shape prior, facilitating more accurate and robust symmetry detection. Experiments demonstrate that the proposed method is capable of detecting reflectional and rotational symmetries accurately, and shows good generality in challenging scenarios, such as objects with heavy occlusion and scanning noise. Moreover, it achieves state-of-the-art performance, improving the F1-score over the existing supervised learning method by 2%-11% on the ShapeNet and ScanNet datasets.

Attention based: modeling human perception of reflectional symmetry in the wild

Beyond Planar Symmetry: Modeling Human Perception of Reflection and Rotation Symmetries in the Wild

SymmetryNet

SRN: Side-output Residual Network for Object Symmetry Detection in the Wild

NeRD: Neural 3D Reflection Symmetry Detector

SRN: Side-Output Residual Network for Object Reflection Symmetry Detection and Beyond

Learning to Detect 3D Symmetry from Single-View RGB-D Images with Weak Supervision

Symmetry Strikes Back: From Single-Image Symmetry Detection to 3D Generation

Symmetry Perception by Deep Networks: Inadequacy of Feed-Forward Architectures and Improvements with Recurrent Connections

Symmetry Awareness Encoded Deep Learning Framework for Brain Imaging Analysis

Reflection Symmetry Axes Localization Framework for Object Intelligent Perception based on Salient Symmetry Feature

An adaptive symmetry detection algorithm based on local features

Reflection Symmetry Axes Localization Framework for Intelligent Object Perception Based on Salient Symmetry Feature

PRS-Net: Planar Reflective Symmetry Detection Net for 3D Models

Learning to Detect 3D Reflection Symmetry for Single-View Reconstruction

Learning-based Real-time Detection of Intrinsic Reflectional Symmetry

Learning-based Intrinsic Reflectional Symmetry Detection

SNeS: Learning Probably Symmetric Neural Surfaces from Incomplete Data

Finding Mirror Symmetry via Registration

Reflection and Rotation Symmetry Detection via Equivariant Learning

Symmetry-Aware Semantic Segmentation of a 3D Human Body Model