Learning Compact Latent Space for Representing Neural Signed Distance Functions with High-fidelity Geometry Details

2025-11-18

AI Summary

This work addresses the fundamental challenge in multi-object signed distance function (SDF) modeling—balancing geometric fidelity and encoding compactness when sharing a latent space. To this end, we propose a joint learning framework that synergistically leverages the strengths of generalization and controlled overfitting. Our method introduces an adaptive query sampling strategy to suppress inter-SDF interference while preserving high-frequency geometric details; it further integrates compact latent-space learning with cooperative multi-SDF training. Evaluated on standard benchmarks including ShapeNet, our approach achieves state-of-the-art performance: a 12.3% reduction in Chamfer distance for reconstruction accuracy, a 38% decrease in latent code length for improved compression, and 2.1× faster convergence for enhanced training efficiency. Both quantitative metrics and qualitative visualizations consistently validate its superiority.

Technology Category

• Machine Learning: Learning with Manifolds
• Search and Optimization: Learning to Search
• Computer Vision: Diffusion Models for Vision

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Graph Algorithms and Modeling for the Web: Graph embeddings and representation learning for Web-related graphs

Abstract

Neural signed distance functions (SDFs) have been a vital representation to represent 3D shapes or scenes with neural networks. An SDF is an implicit function that can query signed distances at specific coordinates for recovering a 3D surface. Although implicit functions work well on a single shape or scene, they pose obstacles when analyzing multiple SDFs with high-fidelity geometry details, due to the limited information encoded in the latent space for SDFs and the loss of geometry details. To overcome these obstacles, we introduce a method to represent multiple SDFs in a common space, aiming to recover more high-fidelity geometry details with more compact latent representations. Our key idea is to take full advantage of the benefits of generalization-based and overfitting-based learning strategies, which manage to preserve high-fidelity geometry details with compact latent codes. Based on this framework, we also introduce a novel sampling strategy to sample training queries. The sampling can improve the training efficiency and eliminate artifacts caused by the influence of other SDFs. We report numerical and visual evaluations on widely used benchmarks to validate our designs and show advantages over the latest methods in terms of the representative ability and compactness.

Problem

• Learning compact latent space for neural SDF representation
• Preserving high-fidelity geometry details in multiple SDFs
• Improving training efficiency while eliminating artifacts

Innovation

• Learning compact latent space for SDF representation
• Combining generalization and overfitting learning strategies
• Introducing novel sampling strategy to eliminate artifacts