RetrievalFuse: Neural 3D Scene Reconstruction with a Database

TL;DR

RetrievalFuse is a neural 3D scene reconstruction method that leverages a scene database for improved detail and coherence, outperforming existing models in accuracy and structure transfer.

Contribution

The paper introduces a novel retrieval-based neural reconstruction approach that combines database-driven scene synthesis with attention-based refinement for enhanced 3D reconstruction.

Findings

Achieves over 8% IoU improvement over state-of-the-art methods.

Effectively transfers local details and coherent structures from training scenes.

Demonstrates superior performance in 3D super resolution and surface reconstruction.

Abstract

3D reconstruction of large scenes is a challenging problem due to the high-complexity nature of the solution space, in particular for generative neural networks. In contrast to traditional generative learned models which encode the full generative process into a neural network and can struggle with maintaining local details at the scene level, we introduce a new method that directly leverages scene geometry from the training database. First, we learn to synthesize an initial estimate for a 3D scene, constructed by retrieving a top-k set of volumetric chunks from the scene database. These candidates are then refined to a final scene generation with an attention-based refinement that can effectively select the most consistent set of geometry from the candidates and combine them together to create an output scene, facilitating transfer of coherent structures and local detail from train scene geometry. We demonstrate our neural scene reconstruction with a database for the tasks of 3D super resolution and surface reconstruction from sparse point clouds, showing that our approach enables generation of more coherent, accurate 3D scenes, improving on average by over 8% in IoU over state-of-the-art scene reconstruction.