Dynamic Scene Understanding through Object-Centric Voxelization and Neural Rendering

TL;DR

DynaVol-S is a novel 3D generative model that captures dynamic scenes with object-centric voxelization and neural rendering, enabling unsupervised scene understanding, semantic segmentation, and scene editing.

Contribution

It introduces a 3D object-centric voxelization framework with semantic integration within a differentiable volume rendering pipeline, advancing dynamic scene modeling and unsupervised decomposition.

Findings

Outperforms existing models in novel view synthesis

Enables unsupervised scene decomposition

Supports scene editing and manipulation

Abstract

Learning object-centric representations from unsupervised videos is challenging. Unlike most previous approaches that focus on decomposing 2D images, we present a 3D generative model named DynaVol-S for dynamic scenes that enables object-centric learning within a differentiable volume rendering framework. The key idea is to perform object-centric voxelization to capture the 3D nature of the scene, which infers per-object occupancy probabilities at individual spatial locations. These voxel features evolve through a canonical-space deformation function and are optimized in an inverse rendering pipeline with a compositional NeRF. Additionally, our approach integrates 2D semantic features to create 3D semantic grids, representing the scene through multiple disentangled voxel grids. DynaVol-S significantly outperforms existing models in both novel view synthesis and unsupervised decomposition tasks for dynamic scenes. By jointly considering geometric structures and semantic features, it effectively addresses challenging real-world scenarios involving complex object interactions. Furthermore, once trained, the explicitly meaningful voxel features enable additional capabilities that 2D scene decomposition methods cannot achieve, such as novel scene generation through editing geometric shapes or manipulating the motion trajectories of objects.