DriveTok: 3D Driving Scene Tokenization for Unified Multi-View Reconstruction and Understanding

TL;DR

DriveTok introduces a novel 3D scene tokenizer for autonomous driving that efficiently unifies multi-view reconstruction and understanding by integrating semantic, geometric, and textural information into scene tokens.

Contribution

It proposes a new 3D driving scene tokenizer that improves multi-view scene understanding and reconstruction using semantic-rich features and deformable cross-attention.

Findings

Outperforms existing methods on nuScenes in reconstruction and segmentation tasks

Achieves efficient multi-view 3D scene understanding with unified tokens

Demonstrates strong performance in depth and occupancy prediction

Abstract

With the growing adoption of vision-language-action models and world models in autonomous driving systems, scalable image tokenization becomes crucial as the interface for the visual modality. However, most existing tokenizers are designed for monocular and 2D scenes, leading to inefficiency and inter-view inconsistency when applied to high-resolution multi-view driving scenes. To address this, we propose DriveTok, an efficient 3D driving scene tokenizer for unified multi-view reconstruction and understanding. DriveTok first obtains semantically rich visual features from vision foundation models and then transforms them into the scene tokens with 3D deformable cross-attention. For decoding, we employ a multi-view transformer to reconstruct multi-view features from the scene tokens and use multiple heads to obtain RGB, depth, and semantic reconstructions. We also add a 3D head directly on the scene tokens for 3D semantic occupancy prediction for better spatial awareness. With the multiple training objectives, DriveTok learns unified scene tokens that integrate semantic, geometric, and textural information for efficient multi-view tokenization. Extensive experiments on the widely used nuScenes dataset demonstrate that the scene tokens from DriveTok perform well on image reconstruction, semantic segmentation, depth prediction, and 3D occupancy prediction tasks.