When Does An Extra View Help? Adapting Single-View 3D Reconstruction with Extra Imagery

2026-08-08

AI Summary

Single-view 3D reconstruction often yields incomplete structures due to limited viewpoint information, and existing methods struggle to effectively leverage additional views to enhance reconstruction quality. This work proposes ASV3D, a framework that dynamically integrates a single auxiliary image during inference without requiring retraining, employing both zero-shot adaptation and contrastive learning–based optimization strategies. ASV3D achieves the first demonstration of cross-view feature alignment without fine-tuning, enhancing multi-view consistency and visual fidelity through contrastive learning. Evaluated on multiple benchmarks and real-world datasets, ASV3D significantly improves reconstruction accuracy and robustness, outperforming state-of-the-art methods in both quantitative metrics and human preference studies.

Abstract

Reconstruction of 3D objects from a single image is a challenging research problem in computer vision. The key challenge is the lack of critical information from viewpoints to complete 3D structures. Using an additional view may help to resolve the issue. However, there is no mechanism that can integrate the extra view into the single-view 3D reconstruction principle. We address this challenge by proposing ASV3D, a framework for adapting single-view 3D object reconstruction to test-time data with support from one additional image. We introduce two adaptation strategies: (i) a zero-shot adaptation scheme that leverages the auxiliary image to improve the reconstruction quality of an object without retraining, and (ii) an optimised adaptation scheme that further enhances visual fidelity and cross-view consistency via contrastive learning. We apply our ASV3D to improve two state-of-the-art single-view 3D reconstruction pipelines on both benchmark and real-world datasets. Results demonstrate that our approach consistently improves reconstruction accuracy and robustness under unconstrained multi-view inputs, outperforming the baselines in both quantitative metrics and human preference. We publish our code and the real-world object dataset in our project page at https://github.com/YNhuHuynh/ASV3D/tree/main.

Problem

• single-view 3D reconstruction
• multi-view integration
• 3D object reconstruction
• viewpoint information
• extra imagery

Innovation

• single-view 3D reconstruction
• test-time adaptation
• multi-view consistency
• contrastive learning
• zero-shot adaptation