HDNET: Exploiting HD Maps for 3D Object Detection

Bin Yang, Ming Liang, Raquel Urtasun
Proceedings of The 2nd Conference on Robot Learning, PMLR 87:146-155, 2018.

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 [1] 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.

Cite this Paper

BibTeX
@InProceedings{pmlr-v87-yang18b, title = {HDNET: Exploiting HD Maps for 3D Object Detection}, author = {Yang, Bin and Liang, Ming and Urtasun, Raquel}, booktitle = {Proceedings of The 2nd Conference on Robot Learning}, pages = {146--155}, year = {2018}, editor = {Billard, Aude and Dragan, Anca and Peters, Jan and Morimoto, Jun}, volume = {87}, series = {Proceedings of Machine Learning Research}, month = {29--31 Oct}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v87/yang18b/yang18b.pdf}, url = {https://proceedings.mlr.press/v87/yang18b.html}, 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 [1] 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. } }
Endnote
%0 Conference Paper %T HDNET: Exploiting HD Maps for 3D Object Detection %A Bin Yang %A Ming Liang %A Raquel Urtasun %B Proceedings of The 2nd Conference on Robot Learning %C Proceedings of Machine Learning Research %D 2018 %E Aude Billard %E Anca Dragan %E Jan Peters %E Jun Morimoto %F pmlr-v87-yang18b %I PMLR %P 146--155 %U https://proceedings.mlr.press/v87/yang18b.html %V 87 %X 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 [1] 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.
APA
Yang, B., Liang, M. & Urtasun, R.. (2018). HDNET: Exploiting HD Maps for 3D Object Detection. Proceedings of The 2nd Conference on Robot Learning, in Proceedings of Machine Learning Research 87:146-155 Available from https://proceedings.mlr.press/v87/yang18b.html.

Related Material