MetaSSP: Enhancing Semi-supervised Implicit 3D Reconstruction through Meta-adaptive EMA and SDF-aware Pseudo-label Evaluation

2026-02-05

AI Summary

This work addresses the challenge of scaling implicit signed distance function (SDF)-based single-view 3D reconstruction to large-scale scenarios, where reliance on extensive labeled data limits practical applicability. To this end, we propose a semi-supervised framework that effectively leverages a small set of annotated images alongside abundant unlabeled data. Our approach introduces two key innovations: a gradient-based parameter importance estimation mechanism that guides exponential moving average (EMA) updates for meta-adaptive consistency, and a pseudo-label weighting strategy that jointly considers augmentation consistency and SDF variance. Evaluated on the Pix3D benchmark, our method achieves a new state of the art, reducing Chamfer Distance by 20.61% and improving Intersection over Union (IoU) by 24.09% compared to existing semi-supervised approaches.

Technology Category

• Machine Learning: Semi-Supervised Learning
• Computer Vision: 3D Computer Vision
• Search and Optimization: Sampling/Simulation-based Search

Application Category

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

Abstract

Implicit SDF-based methods for single-view 3D reconstruction achieve high-quality surfaces but require large labeled datasets, limiting their scalability. We propose MetaSSP, a novel semi-supervised framework that exploits abundant unlabeled images. Our approach introduces gradient-based parameter importance estimation to regularize adaptive EMA updates and an SDF-aware pseudo-label weighting mechanism combining augmentation consistency with SDF variance. Beginning with a 10% supervised warm-up, the unified pipeline jointly refines labeled and unlabeled data. On the Pix3D benchmark, our method reduces Chamfer Distance by approximately 20.61% and increases IoU by around 24.09% compared to existing semi-supervised baselines, setting a new state of the art.

Problem

• 3D reconstruction
• semi-supervised learning
• implicit SDF
• pseudo-labeling
• scalability

Innovation

• Meta-adaptive EMA
• SDF-aware pseudo-labeling
• semi-supervised 3D reconstruction
• implicit SDF
• gradient-based parameter importance