Weakly-Supervised 3D Visual Grounding based on Visual Language Alignment

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Xu, Xiaoxu, Yuan, Yitian, Zhang, Qiudan, Wu, Wenhui, Jie, Zequn, Ma, Lin, Wang, Xu
Format: Preprint
Published: 2023
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866918130321719296
author Xu, Xiaoxu
Yuan, Yitian
Zhang, Qiudan
Wu, Wenhui
Jie, Zequn
Ma, Lin
Wang, Xu
author_facet Xu, Xiaoxu
Yuan, Yitian
Zhang, Qiudan
Wu, Wenhui
Jie, Zequn
Ma, Lin
Wang, Xu
contents Learning to ground natural language queries to target objects or regions in 3D point clouds is quite essential for 3D scene understanding. Nevertheless, existing 3D visual grounding approaches require a substantial number of bounding box annotations for text queries, which is time-consuming and labor-intensive to obtain. In this paper, we propose 3D-VLA, a weakly supervised approach for 3D visual grounding based on Visual Linguistic Alignment. Our 3D-VLA exploits the superior ability of current large-scale vision-language models (VLMs) on aligning the semantics between texts and 2D images, as well as the naturally existing correspondences between 2D images and 3D point clouds, and thus implicitly constructs correspondences between texts and 3D point clouds with no need for fine-grained box annotations in the training procedure. During the inference stage, the learned text-3D correspondence will help us ground the text queries to the 3D target objects even without 2D images. To the best of our knowledge, this is the first work to investigate 3D visual grounding in a weakly supervised manner by involving large scale vision-language models, and extensive experiments on ReferIt3D and ScanRefer datasets demonstrate that our 3D-VLA achieves comparable and even superior results over the fully supervised methods.
format Preprint
id arxiv_https___arxiv_org_abs_2312_09625
institution arXiv
publishDate 2023
record_format arxiv
spellingShingle Weakly-Supervised 3D Visual Grounding based on Visual Language Alignment
Xu, Xiaoxu
Yuan, Yitian
Zhang, Qiudan
Wu, Wenhui
Jie, Zequn
Ma, Lin
Wang, Xu
Computer Vision and Pattern Recognition
Computation and Language
Learning to ground natural language queries to target objects or regions in 3D point clouds is quite essential for 3D scene understanding. Nevertheless, existing 3D visual grounding approaches require a substantial number of bounding box annotations for text queries, which is time-consuming and labor-intensive to obtain. In this paper, we propose 3D-VLA, a weakly supervised approach for 3D visual grounding based on Visual Linguistic Alignment. Our 3D-VLA exploits the superior ability of current large-scale vision-language models (VLMs) on aligning the semantics between texts and 2D images, as well as the naturally existing correspondences between 2D images and 3D point clouds, and thus implicitly constructs correspondences between texts and 3D point clouds with no need for fine-grained box annotations in the training procedure. During the inference stage, the learned text-3D correspondence will help us ground the text queries to the 3D target objects even without 2D images. To the best of our knowledge, this is the first work to investigate 3D visual grounding in a weakly supervised manner by involving large scale vision-language models, and extensive experiments on ReferIt3D and ScanRefer datasets demonstrate that our 3D-VLA achieves comparable and even superior results over the fully supervised methods.
title Weakly-Supervised 3D Visual Grounding based on Visual Language Alignment
topic Computer Vision and Pattern Recognition
Computation and Language
url https://arxiv.org/abs/2312.09625