FastMap: Fast Queries Initialization Based Vectorized HD Map Reconstruction Framework

2025-03-07

AI Summary

To address the low computational efficiency of DETR-based vectorized map reconstruction methods caused by multi-layer decoders, this paper proposes a single-layer, two-stage Transformer architecture. Our key contributions are: (1) a novel heatmap-guided structured query generation module that replaces random initialization with geometry-aware, learnable query initialization; (2) a geometrically constrained point-to-line loss to enhance discriminability among homogeneous features; and (3) integration of learnable positional encodings to improve multi-level representation efficiency. Evaluated on nuScenes and Argoverse2, our method achieves state-of-the-art performance while accelerating decoder inference by 3.2×. It thus strikes a strong balance between accuracy and real-time capability, establishing an efficient new paradigm for online high-definition map vectorization in autonomous driving.

Technology Category

Search and Optimization: Learning to SearchComputer Vision: Representation Learning for VisionPlanning, Routing, and Scheduling: Planning with Language Models

Application Category

Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for rankingGraph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphsSystems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications

Abstract

Reconstruction of high-definition maps is a crucial task in perceiving the autonomous driving environment, as its accuracy directly impacts the reliability of prediction and planning capabilities in downstream modules. Current vectorized map reconstruction methods based on the DETR framework encounter limitations due to the redundancy in the decoder structure, necessitating the stacking of six decoder layers to maintain performance, which significantly hampers computational efficiency. To tackle this issue, we introduce FastMap, an innovative framework designed to reduce decoder redundancy in existing approaches. FastMap optimizes the decoder architecture by employing a single-layer, two-stage transformer that achieves multilevel representation capabilities. Our framework eliminates the conventional practice of randomly initializing queries and instead incorporates a heatmap-guided query generation module during the decoding phase, which effectively maps image features into structured query vectors using learnable positional encoding. Additionally, we propose a geometry-constrained point-to-line loss mechanism for FastMap, which adeptly addresses the challenge of distinguishing highly homogeneous features that often arise in traditional point-to-point loss computations. Extensive experiments demonstrate that FastMap achieves state-of-the-art performance in both nuScenes and Argoverse2 datasets, with its decoder operating 3.2 faster than the baseline. Code and more demos are available at https://github.com/hht1996ok/FastMap.

Problem

• Reduces decoder redundancy in HD map reconstruction
• Improves computational efficiency with single-layer transformer
• Enhances feature distinction with geometry-constrained loss

Innovation

• Single-layer two-stage transformer optimization
• Heatmap-guided query generation module
• Geometry-constrained point-to-line loss mechanism