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

TL;DR

MetaSSP is a semi-supervised framework that improves implicit 3D reconstruction from single images by leveraging unlabeled data with novel regularization and pseudo-label evaluation techniques, achieving state-of-the-art results.

Contribution

The paper introduces MetaSSP, a semi-supervised method with gradient-based importance estimation and SDF-aware pseudo-label weighting for 3D reconstruction.

Findings

Reduces Chamfer Distance by ~20.61% on Pix3D

Increases IoU by ~24.09% on Pix3D

Sets new state-of-the-art performance in semi-supervised 3D reconstruction

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.