Cargando…
One-Stage Anchor-Free 3D Vehicle Detection from LiDAR Sensors
Recent one-stage 3D detection methods generate anchor boxes with various sizes and orientations in the ground plane, then determine whether these anchor boxes contain any region of interest and adjust the edges of them for accurate object bounding boxes. The anchor-based algorithm calculates the cla...
Autores principales: | , , , , |
---|---|
Formato: | Online Artículo Texto |
Lenguaje: | English |
Publicado: |
MDPI
2021
|
Materias: | |
Acceso en línea: | https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8069010/ https://www.ncbi.nlm.nih.gov/pubmed/33918952 http://dx.doi.org/10.3390/s21082651 |
_version_ | 1783683136938311680 |
---|---|
author | Li, Hao Zhao, Sanyuan Zhao, Wenjun Zhang, Libin Shen, Jianbing |
author_facet | Li, Hao Zhao, Sanyuan Zhao, Wenjun Zhang, Libin Shen, Jianbing |
author_sort | Li, Hao |
collection | PubMed |
description | Recent one-stage 3D detection methods generate anchor boxes with various sizes and orientations in the ground plane, then determine whether these anchor boxes contain any region of interest and adjust the edges of them for accurate object bounding boxes. The anchor-based algorithm calculates the classification and regression label for each anchor box during the training process, which is inefficient and complicated. We propose a one-stage, anchor-free 3D vehicle detection algorithm based on LiDAR point clouds. The object position is encoded as a set of keypoints in the bird’s-eye view (BEV) of point clouds. We apply the voxel/pillar feature extractor and convolutional blocks to map an unstructured point cloud to a single-channel 2D heatmap. The vehicle’s Z-axis position, dimension, and orientation angle are regressed as additional attributes of the keypoints. Our method combines SmoothL1 loss and IoU (Intersection over Union) loss, and we apply [Formula: see text] as angle regression labels, which achieve high average orientation similarity (AOS) without any direction classification tricks. During the target assignment and bounding box decoding process, our framework completely avoids any calculations related to anchor boxes. Our framework is end-to-end training and stands at the same performance level as the other one-stage anchor-based detectors. |
format | Online Article Text |
id | pubmed-8069010 |
institution | National Center for Biotechnology Information |
language | English |
publishDate | 2021 |
publisher | MDPI |
record_format | MEDLINE/PubMed |
spelling | pubmed-80690102021-04-26 One-Stage Anchor-Free 3D Vehicle Detection from LiDAR Sensors Li, Hao Zhao, Sanyuan Zhao, Wenjun Zhang, Libin Shen, Jianbing Sensors (Basel) Article Recent one-stage 3D detection methods generate anchor boxes with various sizes and orientations in the ground plane, then determine whether these anchor boxes contain any region of interest and adjust the edges of them for accurate object bounding boxes. The anchor-based algorithm calculates the classification and regression label for each anchor box during the training process, which is inefficient and complicated. We propose a one-stage, anchor-free 3D vehicle detection algorithm based on LiDAR point clouds. The object position is encoded as a set of keypoints in the bird’s-eye view (BEV) of point clouds. We apply the voxel/pillar feature extractor and convolutional blocks to map an unstructured point cloud to a single-channel 2D heatmap. The vehicle’s Z-axis position, dimension, and orientation angle are regressed as additional attributes of the keypoints. Our method combines SmoothL1 loss and IoU (Intersection over Union) loss, and we apply [Formula: see text] as angle regression labels, which achieve high average orientation similarity (AOS) without any direction classification tricks. During the target assignment and bounding box decoding process, our framework completely avoids any calculations related to anchor boxes. Our framework is end-to-end training and stands at the same performance level as the other one-stage anchor-based detectors. MDPI 2021-04-09 /pmc/articles/PMC8069010/ /pubmed/33918952 http://dx.doi.org/10.3390/s21082651 Text en © 2021 by the authors. https://creativecommons.org/licenses/by/4.0/Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/). |
spellingShingle | Article Li, Hao Zhao, Sanyuan Zhao, Wenjun Zhang, Libin Shen, Jianbing One-Stage Anchor-Free 3D Vehicle Detection from LiDAR Sensors |
title | One-Stage Anchor-Free 3D Vehicle Detection from LiDAR Sensors |
title_full | One-Stage Anchor-Free 3D Vehicle Detection from LiDAR Sensors |
title_fullStr | One-Stage Anchor-Free 3D Vehicle Detection from LiDAR Sensors |
title_full_unstemmed | One-Stage Anchor-Free 3D Vehicle Detection from LiDAR Sensors |
title_short | One-Stage Anchor-Free 3D Vehicle Detection from LiDAR Sensors |
title_sort | one-stage anchor-free 3d vehicle detection from lidar sensors |
topic | Article |
url | https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8069010/ https://www.ncbi.nlm.nih.gov/pubmed/33918952 http://dx.doi.org/10.3390/s21082651 |
work_keys_str_mv | AT lihao onestageanchorfree3dvehicledetectionfromlidarsensors AT zhaosanyuan onestageanchorfree3dvehicledetectionfromlidarsensors AT zhaowenjun onestageanchorfree3dvehicledetectionfromlidarsensors AT zhanglibin onestageanchorfree3dvehicledetectionfromlidarsensors AT shenjianbing onestageanchorfree3dvehicledetectionfromlidarsensors |