OVIR-3D: Open-Vocabulary 3D Instance Retrieval Without Training on 3D Data

TL;DR

OVIR-3D introduces a real-time, training-free method for open-vocabulary 3D object retrieval using multi-view 2D proposals fused into 3D space, leveraging 2D datasets for effective retrieval.

Contribution

The paper proposes a novel approach that performs open-vocabulary 3D instance retrieval without any 3D training data, utilizing multi-view 2D proposals fused into 3D.

Findings

Effective retrieval on public datasets and real robot scenarios.

Real-time performance in indoor scenes.

No additional 3D training required.

Abstract

This work presents OVIR-3D, a straightforward yet effective method for open-vocabulary 3D object instance retrieval without using any 3D data for training. Given a language query, the proposed method is able to return a ranked set of 3D object instance segments based on the feature similarity of the instance and the text query. This is achieved by a multi-view fusion of text-aligned 2D region proposals into 3D space, where the 2D region proposal network could leverage 2D datasets, which are more accessible and typically larger than 3D datasets. The proposed fusion process is efficient as it can be performed in real-time for most indoor 3D scenes and does not require additional training in 3D space. Experiments on public datasets and a real robot show the effectiveness of the method and its potential for applications in robot navigation and manipulation.