Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Distribution Guidance Network for Weakly Supervised Point Cloud Semantic Segmentation

Zhiyi Pan, Wei Gao, Shan Liu, Ge Li

TL;DR

This work develops a Distribution Guidance Network (DGNet), which comprises a weakly supervised learning branch and a distribution alignment branch that alternately updates the parameters of the moVMF and the network, ensuring alignment with the moVMF-defined latent space.

Abstract

Despite alleviating the dependence on dense annotations inherent to fully supervised methods, weakly supervised point cloud semantic segmentation suffers from inadequate supervision signals. In response to this challenge, we introduce a novel perspective that imparts auxiliary constraints by regulating the feature space under weak supervision. Our initial investigation identifies which distributions accurately characterize the feature space, subsequently leveraging this priori to guide the alignment of the weakly supervised embeddings. Specifically, we analyze the superiority of the mixture of von Mises-Fisher distributions (moVMF) among several common distribution candidates. Accordingly, we develop a Distribution Guidance Network (DGNet), which comprises a weakly supervised learning branch and a distribution alignment branch. Leveraging reliable clustering initialization derived from the weakly supervised learning branch, the distribution alignment branch alternately updates the parameters of the moVMF and the network, ensuring alignment with the moVMF-defined latent space. Extensive experiments validate the rationality and effectiveness of our distribution choice and network design. Consequently, DGNet achieves state-of-the-art performance under multiple datasets and various weakly supervised settings.

Distribution Guidance Network for Weakly Supervised Point Cloud Semantic Segmentation

TL;DR

This work develops a Distribution Guidance Network (DGNet), which comprises a weakly supervised learning branch and a distribution alignment branch that alternately updates the parameters of the moVMF and the network, ensuring alignment with the moVMF-defined latent space.

Abstract

Despite alleviating the dependence on dense annotations inherent to fully supervised methods, weakly supervised point cloud semantic segmentation suffers from inadequate supervision signals. In response to this challenge, we introduce a novel perspective that imparts auxiliary constraints by regulating the feature space under weak supervision. Our initial investigation identifies which distributions accurately characterize the feature space, subsequently leveraging this priori to guide the alignment of the weakly supervised embeddings. Specifically, we analyze the superiority of the mixture of von Mises-Fisher distributions (moVMF) among several common distribution candidates. Accordingly, we develop a Distribution Guidance Network (DGNet), which comprises a weakly supervised learning branch and a distribution alignment branch. Leveraging reliable clustering initialization derived from the weakly supervised learning branch, the distribution alignment branch alternately updates the parameters of the moVMF and the network, ensuring alignment with the moVMF-defined latent space. Extensive experiments validate the rationality and effectiveness of our distribution choice and network design. Consequently, DGNet achieves state-of-the-art performance under multiple datasets and various weakly supervised settings.

Paper Structure

This paper contains 21 sections, 10 equations, 4 figures, 9 tables, 1 algorithm.

Table of Contents

  1. Introduction
  2. Related Work
  3. Weakly Supervised Point Cloud Semantic Segmentation.
  4. Feature Distribution Constraints.
  5. Mixture of von Mises-Fisher Distributions.
  6. Methodology
  7. Preliminaries
  8. Task Definition.
  9. von Mises-Fisher Distribution (vMF).
  10. Mixture of vMF (moVMF).
  11. Feature Space Description
  12. Distribution Guidance Network
  13. Weakly Supervised Learning Branch
  14. Distribution Alignment Branch
  15. Loss Function
  16. ...and 6 more sections

Figures (4)

  • Figure 1: Visual comparisons of mainstream weakly supervised point cloud semantic segmentation paradigms and our DGNet. The solid and dashed lines represent the network forward process and the loss function, respectively.
  • Figure 2: Structure of Distribution Guidance Network.
  • Figure 3: DGNet provides segmentation predictions from the weakly supervised learning branch and explains it probabilistically by posterior probabilities from the distribution alignment branch.
  • Figure 4: Visual comparisons between baseline and our DGNet on S3DIS Area 5 at 0.01% label rate.