EFEM: Equivariant Neural Field Expectation Maximization
for 3D Object Segmentation Without Scene Supervision

Jiahui Lei1        Congyue Deng2        Karl Schmeckpeper1        Leonidas Guibas2        Kostas Daniilidis1

Abstraction

We introduce Equivariant Neural Field Expectation Maximization (EFEM), a simple, effective, and robust geometric algorithm that can segment objects in 3D scenes without annotations or training on scenes. We achieve such unsupervised segmentation by exploiting single object shape priors. We make two novel steps in that direc-tion. First, we introduce equivariant shape representations to this problem to eliminate the complexity induced by the variation in object configuration. Second, we propose a novel EM algorithm that can iteratively refine segmentation masks using the equivariant shape prior. We collect a novel real dataset Chairs and Mugs that contains various object configurations and novel scenes in order to verify the effectiveness and robustness of our method. Experimental results demonstrate that our method achieves consistent and robust performance across different scenes where the (weakly) supervised methods may fail.

Introduction

Watch this video here on Bilibili in China

Train Equivariant Single Object Shape Prior on ShapeNet

Inference on Real Scenes

Results

Citation

@inproceedings{Lei2023EFEM,
title={EFEM: Equivariant Neural Field Expectation Maximization for 3D Object Segmentation Without Scene Supervision},
author={Lei, Jiahui and Deng, Congyue and Schmeckpeper, Karl and Guibas, Leonidas and Daniilidis, Kostas},
booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
url={https://cis.upenn.edu/~leijh/projects/efem},
year={2023}
}