HDNET: Exploiting HD Maps for 3D Object Detection

TL;DR

This paper introduces HDNET, a 3D object detection framework that leverages HD maps for improved accuracy and robustness, incorporating a map prediction module for scenarios lacking map data, and achieves real-time performance.

Contribution

HDNET is the first single-stage detector to utilize HD maps and a map prediction module for enhanced 3D detection in both mapped and un-mapped environments.

Findings

Outperforms state-of-the-art methods on KITTI and large-scale benchmarks.

Operates at 20 frames per second, enabling real-time applications.

Effectively handles scenarios without HD maps through map prediction.

Abstract

In this paper we show that High-Definition (HD) maps provide strong priors that can boost the performance and robustness of modern 3D object detectors. Towards this goal, we design a single stage detector that extracts geometric and semantic features from the HD maps. As maps might not be available everywhere, we also propose a map prediction module that estimates the map on the fly from raw LiDAR data. We conduct extensive experiments on KITTI as well as a large-scale 3D detection benchmark containing 1 million frames, and show that the proposed map-aware detector consistently outperforms the state-of-the-art in both mapped and un-mapped scenarios. Importantly the whole framework runs at 20 frames per second.