Search3D: Hierarchical Open-Vocabulary 3D Segmentation

TL;DR

Search3D introduces a hierarchical open-vocabulary 3D scene representation that enables flexible search at multiple levels of detail, including parts, objects, and attributes, advancing 3D scene understanding.

Contribution

It presents Search3D, a novel method for hierarchical open-vocabulary 3D segmentation, and provides a new benchmark with annotations for evaluating fine-grained scene components.

Findings

Outperforms baselines in 3D part segmentation

Effective in segmenting objects and materials

Supports multi-level scene search

Abstract

Open-vocabulary 3D segmentation enables exploration of 3D spaces using free-form text descriptions. Existing methods for open-vocabulary 3D instance segmentation primarily focus on identifying object-level instances but struggle with finer-grained scene entities such as object parts, or regions described by generic attributes. In this work, we introduce Search3D, an approach to construct hierarchical open-vocabulary 3D scene representations, enabling 3D search at multiple levels of granularity: fine-grained object parts, entire objects, or regions described by attributes like materials. Unlike prior methods, Search3D shifts towards a more flexible open-vocabulary 3D search paradigm, moving beyond explicit object-centric queries. For systematic evaluation, we further contribute a scene-scale open-vocabulary 3D part segmentation benchmark based on MultiScan, along with a set of open-vocabulary fine-grained part annotations on ScanNet++. Search3D outperforms baselines in scene-scale open-vocabulary 3D part segmentation, while maintaining strong performance in segmenting 3D objects and materials. Our project page is http://search3d-segmentation.github.io.