Learning Object-Compositional Neural Radiance Field for Editable Scene Rendering

TL;DR

This paper introduces an object-compositional neural radiance field system that enables realistic scene rendering and high-level object editing in cluttered scenes, overcoming limitations of existing scene-agnostic methods.

Contribution

It proposes a novel two-pathway neural architecture with scene and object branches, and a scene-guided training strategy for cluttered scenes, enhancing object-level editing capabilities.

Findings

Achieves competitive novel-view synthesis performance.

Enables realistic object-level editing in complex scenes.

Effectively handles occlusions and object boundaries.

Abstract

Implicit neural rendering techniques have shown promising results for novel view synthesis. However, existing methods usually encode the entire scene as a whole, which is generally not aware of the object identity and limits the ability to the high-level editing tasks such as moving or adding furniture. In this paper, we present a novel neural scene rendering system, which learns an object-compositional neural radiance field and produces realistic rendering with editing capability for a clustered and real-world scene. Specifically, we design a novel two-pathway architecture, in which the scene branch encodes the scene geometry and appearance, and the object branch encodes each standalone object conditioned on learnable object activation codes. To survive the training in heavily cluttered scenes, we propose a scene-guided training strategy to solve the 3D space ambiguity in the occluded regions and learn sharp boundaries for each object. Extensive experiments demonstrate that our system not only achieves competitive performance for static scene novel-view synthesis, but also produces realistic rendering for object-level editing.