GeoFusionLRM: Geometry-Aware Self-Correction for Consistent 3D Reconstruction

2026-02-15

AI Summary

Single-image 3D reconstruction often suffers from reduced fidelity due to geometric inconsistencies and misaligned details. To address this, this work proposes a geometry-aware self-correction framework that, for the first time, integrates self-generated normals and depth as geometric priors within large reconstruction models (LRMs). The approach employs a Transformer-based feedback mechanism coupled with a feature fusion module to enable closed-loop refinement, operating entirely without additional supervision or external signals. This method significantly enhances the consistency between reconstructed geometry and the input view, outperforming state-of-the-art techniques in terms of geometric sharpness, normal orientation coherence, and overall reconstruction fidelity.

Technology Category

• Computer Vision: Large Vision Models
• Machine Learning: Large Multimodal Models (LMMs)
• Intelligent Robots: Multimodal Perception & Sensor Fusion

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Single-image 3D reconstruction with large reconstruction models (LRMs) has advanced rapidly, yet reconstructions often exhibit geometric inconsistencies and misaligned details that limit fidelity. We introduce GeoFusionLRM, a geometry-aware self-correction framework that leverages the model's own normal and depth predictions to refine structural accuracy. Unlike prior approaches that rely solely on features extracted from the input image, GeoFusionLRM feeds back geometric cues through a dedicated transformer and fusion module, enabling the model to correct errors and enforce consistency with the conditioning image. This design improves the alignment between the reconstructed mesh and the input views without additional supervision or external signals. Extensive experiments demonstrate that GeoFusionLRM achieves sharper geometry, more consistent normals, and higher fidelity than state-of-the-art LRM baselines.

Problem

• 3D reconstruction
• geometric inconsistency
• single-image
• fidelity
• large reconstruction models

Innovation

• geometry-aware self-correction
• 3D reconstruction
• large reconstruction models
• normal-depth feedback
• consistency enforcement