GENERATING SYNTHETIC 3D POINT SEGMENTS FOR IMPROVED CLASSIFICATION OF MOBILE LIDAR POINT CLOUDS
Mobile lidar point clouds are commonly used for 3d mapping of road environments as they provide a rich, highly detailed geometric representation of objects on and around the road. However, raw lidar point clouds lack semantic information about the type of objects, which is necessary for various appl...
Main Authors: | , , |
---|---|
Format: | Article |
Language: | English |
Published: |
Copernicus Publications
2021-06-01
|
Series: | The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences |
Online Access: | https://www.int-arch-photogramm-remote-sens-spatial-inf-sci.net/XLIII-B2-2021/139/2021/isprs-archives-XLIII-B2-2021-139-2021.pdf |
id |
doaj-14c80ad3146c4fd88abfd082c3b05fae |
---|---|
record_format |
Article |
spelling |
doaj-14c80ad3146c4fd88abfd082c3b05fae2021-06-28T22:22:22ZengCopernicus PublicationsThe International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences1682-17502194-90342021-06-01XLIII-B2-202113914410.5194/isprs-archives-XLIII-B2-2021-139-2021GENERATING SYNTHETIC 3D POINT SEGMENTS FOR IMPROVED CLASSIFICATION OF MOBILE LIDAR POINT CLOUDSS. A. Chitnis0Z. Huang1K. Khoshelham2University of Melbourne, Parkville, Victoria 3010, AustraliaUniversity of Melbourne, Parkville, Victoria 3010, AustraliaUniversity of Melbourne, Parkville, Victoria 3010, AustraliaMobile lidar point clouds are commonly used for 3d mapping of road environments as they provide a rich, highly detailed geometric representation of objects on and around the road. However, raw lidar point clouds lack semantic information about the type of objects, which is necessary for various applications. Existing methods for the classification of objects in mobile lidar data, including state of the art deep learning methods, achieve relatively low accuracies, and a primary reason for this under-performance is the inadequacy of available 3d training samples to sufficiently train deep networks. In this paper, we propose a generative model for creating synthetic 3d point segments that can aid in improving the classification performance of mobile lidar point clouds. We train a 3d Adversarial Autoencoder (3dAAE) to generate synthetic point segments that exhibit a high resemblance to and share similar geometric features with real point segments. We evaluate the performance of a PointNet-like classifier trained with and without the synthetic point segments. The evaluation results support our hypothesis that training a classifier with training data augmented with synthetic samples leads to significant improvement in the classification performance. Specifically, our model achieves an F1 score of 0.94 for vehicles and pedestrians and 1.00 for traffic signs.https://www.int-arch-photogramm-remote-sens-spatial-inf-sci.net/XLIII-B2-2021/139/2021/isprs-archives-XLIII-B2-2021-139-2021.pdf |
collection |
DOAJ |
language |
English |
format |
Article |
sources |
DOAJ |
author |
S. A. Chitnis Z. Huang K. Khoshelham |
spellingShingle |
S. A. Chitnis Z. Huang K. Khoshelham GENERATING SYNTHETIC 3D POINT SEGMENTS FOR IMPROVED CLASSIFICATION OF MOBILE LIDAR POINT CLOUDS The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences |
author_facet |
S. A. Chitnis Z. Huang K. Khoshelham |
author_sort |
S. A. Chitnis |
title |
GENERATING SYNTHETIC 3D POINT SEGMENTS FOR IMPROVED CLASSIFICATION OF MOBILE LIDAR POINT CLOUDS |
title_short |
GENERATING SYNTHETIC 3D POINT SEGMENTS FOR IMPROVED CLASSIFICATION OF MOBILE LIDAR POINT CLOUDS |
title_full |
GENERATING SYNTHETIC 3D POINT SEGMENTS FOR IMPROVED CLASSIFICATION OF MOBILE LIDAR POINT CLOUDS |
title_fullStr |
GENERATING SYNTHETIC 3D POINT SEGMENTS FOR IMPROVED CLASSIFICATION OF MOBILE LIDAR POINT CLOUDS |
title_full_unstemmed |
GENERATING SYNTHETIC 3D POINT SEGMENTS FOR IMPROVED CLASSIFICATION OF MOBILE LIDAR POINT CLOUDS |
title_sort |
generating synthetic 3d point segments for improved classification of mobile lidar point clouds |
publisher |
Copernicus Publications |
series |
The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences |
issn |
1682-1750 2194-9034 |
publishDate |
2021-06-01 |
description |
Mobile lidar point clouds are commonly used for 3d mapping of road environments as they provide a rich, highly detailed geometric representation of objects on and around the road. However, raw lidar point clouds lack semantic information about the type of objects, which is necessary for various applications. Existing methods for the classification of objects in mobile lidar data, including state of the art deep learning methods, achieve relatively low accuracies, and a primary reason for this under-performance is the inadequacy of available 3d training samples to sufficiently train deep networks. In this paper, we propose a generative model for creating synthetic 3d point segments that can aid in improving the classification performance of mobile lidar point clouds. We train a 3d Adversarial Autoencoder (3dAAE) to generate synthetic point segments that exhibit a high resemblance to and share similar geometric features with real point segments. We evaluate the performance of a PointNet-like classifier trained with and without the synthetic point segments. The evaluation results support our hypothesis that training a classifier with training data augmented with synthetic samples leads to significant improvement in the classification performance. Specifically, our model achieves an F1 score of 0.94 for vehicles and pedestrians and 1.00 for traffic signs. |
url |
https://www.int-arch-photogramm-remote-sens-spatial-inf-sci.net/XLIII-B2-2021/139/2021/isprs-archives-XLIII-B2-2021-139-2021.pdf |
work_keys_str_mv |
AT sachitnis generatingsynthetic3dpointsegmentsforimprovedclassificationofmobilelidarpointclouds AT zhuang generatingsynthetic3dpointsegmentsforimprovedclassificationofmobilelidarpointclouds AT kkhoshelham generatingsynthetic3dpointsegmentsforimprovedclassificationofmobilelidarpointclouds |
_version_ |
1721355839673991168 |