CLIP-Guided Vision-Language Pre-training for Question Answering in 3D Scenes

TL;DR

This paper introduces a novel 3D vision-language pre-training method that leverages CLIP to improve 3D scene understanding and question answering, outperforming existing approaches.

Contribution

We propose a CLIP-guided pre-training approach that aligns 3D scene features with 2D image and text embeddings for enhanced 3D reasoning.

Findings

Outperforms state-of-the-art in 3D visual question answering

Produces interpretable 3D scene representations

Demonstrates effective transfer of CLIP's knowledge to 3D tasks

Abstract

Training models to apply linguistic knowledge and visual concepts from 2D images to 3D world understanding is a promising direction that researchers have only recently started to explore. In this work, we design a novel 3D pre-training Vision-Language method that helps a model learn semantically meaningful and transferable 3D scene point cloud representations. We inject the representational power of the popular CLIP model into our 3D encoder by aligning the encoded 3D scene features with the corresponding 2D image and text embeddings produced by CLIP. To assess our model's 3D world reasoning capability, we evaluate it on the downstream task of 3D Visual Question Answering. Experimental quantitative and qualitative results show that our pre-training method outperforms state-of-the-art works in this task and leads to an interpretable representation of 3D scene features.