HDNET: Exploiting HD Maps for 3D Object Detection
arXiv:2012.11704
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.
Spotlight paper at 2nd Conference on Robot Learning (CoRL 2018)
Cited by in corpus (26)
- Argoverse: 3D Tracking and Forecasting with Rich Maps
- Semantics for Robotic Mapping, Perception and Interaction: A Survey
- End-to-End Multi-View Fusion for 3D Object Detection in LiDAR Point Clouds
- Spatially-Aware Graph Neural Networks for Relational Behavior Forecasting from Sensor Data
- RefinedMPL: Refined Monocular PseudoLiDAR for 3D Object Detection in Autonomous Driving
- Learning Lane Graph Representations for Motion Forecasting
- 3D IoU-Net: IoU Guided 3D Object Detector for Point Clouds
- TANet: Robust 3D Object Detection from Point Clouds with Triple Attention
- Pillar-based Object Detection for Autonomous Driving
- Auto4D: Learning to Label 4D Objects from Sequential Point Clouds
- Identifying Unknown Instances for Autonomous Driving
- Associate-3Ddet: Perceptual-to-Conceptual Association for 3D Point Cloud Object Detection
- Discrete Residual Flow for Probabilistic Pedestrian Behavior Prediction
- Depth-conditioned Dynamic Message Propagation for Monocular 3D Object Detection
- SSN: Shape Signature Networks for Multi-class Object Detection from Point Clouds
- End-to-End Pseudo-LiDAR for Image-Based 3D Object Detection
- Multi-Task Multi-Sensor Fusion for 3D Object Detection
- 3D for Free: Crossmodal Transfer Learning using HD Maps
- PnPNet: End-to-End Perception and Prediction with Tracking in the Loop
- S-AT GCN: Spatial-Attention Graph Convolution Network based Feature Enhancement for 3D Object Detection
- YOLOStereo3D: A Step Back to 2D for Efficient Stereo 3D Detection
- Streaming Object Detection for 3-D Point Clouds
- Pseudo-labeling for Scalable 3D Object Detection
- Review: deep learning on 3D point clouds
- Safety-Oriented Pedestrian Motion and Scene Occupancy Forecasting
- DAGMapper: Learning to Map by Discovering Lane Topology