Scene flow estimation from 3D point clouds based on dual‐branch implicit neural representations
Abstract Recently, online optimisation‐based scene flow estimation has attracted significant attention due to its strong domain adaptivity. Although online optimisation‐based methods have made significant advances, the performance is far from satisfactory as only flow priors are considered, neglecti...
Main Authors: | , , , |
---|---|
Format: | Article |
Language: | English |
Published: |
Wiley
2024-03-01
|
Series: | IET Computer Vision |
Subjects: | |
Online Access: | https://doi.org/10.1049/cvi2.12237 |
_version_ | 1797260115802324992 |
---|---|
author | Mingliang Zhai Kang Ni Jiucheng Xie Hao Gao |
author_facet | Mingliang Zhai Kang Ni Jiucheng Xie Hao Gao |
author_sort | Mingliang Zhai |
collection | DOAJ |
description | Abstract Recently, online optimisation‐based scene flow estimation has attracted significant attention due to its strong domain adaptivity. Although online optimisation‐based methods have made significant advances, the performance is far from satisfactory as only flow priors are considered, neglecting scene priors that are crucial for the representations of dynamic scenes. To address this problem, the authors introduce a dual‐branch MLP‐based architecture to encode implicit scene representations from a source 3D point cloud, which can additionally synthesise a target 3D point cloud. Thus, the mapping function between the source and synthesised target 3D point clouds is established as an extra implicit regulariser to capture scene priors. Moreover, their model infers both flow and scene priors in a stronger bidirectional manner. It can effectively establish spatiotemporal constraints among the synthesised, source, and target 3D point clouds. Experiments on four challenging datasets, including KITTI scene flow, FlyingThings3D, Argoverse, and nuScenes, show that our method can achieve potential and comparable results, proving its effectiveness and generality. |
first_indexed | 2024-04-24T23:20:12Z |
format | Article |
id | doaj.art-176df197ad944b508506a4aeaf3abea2 |
institution | Directory Open Access Journal |
issn | 1751-9632 1751-9640 |
language | English |
last_indexed | 2024-04-24T23:20:12Z |
publishDate | 2024-03-01 |
publisher | Wiley |
record_format | Article |
series | IET Computer Vision |
spelling | doaj.art-176df197ad944b508506a4aeaf3abea22024-03-16T07:56:04ZengWileyIET Computer Vision1751-96321751-96402024-03-0118221022310.1049/cvi2.12237Scene flow estimation from 3D point clouds based on dual‐branch implicit neural representationsMingliang Zhai0Kang Ni1Jiucheng Xie2Hao Gao3School of Automation Nanjing University of Posts and Telecommunications Nanjing Jiangsu Province ChinaSchool of Computer Science Nanjing University of Posts and Telecommunications Nanjing Jiangsu Province ChinaSchool of Automation Nanjing University of Posts and Telecommunications Nanjing Jiangsu Province ChinaSchool of Automation Nanjing University of Posts and Telecommunications Nanjing Jiangsu Province ChinaAbstract Recently, online optimisation‐based scene flow estimation has attracted significant attention due to its strong domain adaptivity. Although online optimisation‐based methods have made significant advances, the performance is far from satisfactory as only flow priors are considered, neglecting scene priors that are crucial for the representations of dynamic scenes. To address this problem, the authors introduce a dual‐branch MLP‐based architecture to encode implicit scene representations from a source 3D point cloud, which can additionally synthesise a target 3D point cloud. Thus, the mapping function between the source and synthesised target 3D point clouds is established as an extra implicit regulariser to capture scene priors. Moreover, their model infers both flow and scene priors in a stronger bidirectional manner. It can effectively establish spatiotemporal constraints among the synthesised, source, and target 3D point clouds. Experiments on four challenging datasets, including KITTI scene flow, FlyingThings3D, Argoverse, and nuScenes, show that our method can achieve potential and comparable results, proving its effectiveness and generality.https://doi.org/10.1049/cvi2.12237image enhancementimage motion analysisimage sensorslearning (artificial intelligence)motion estimationobject detection |
spellingShingle | Mingliang Zhai Kang Ni Jiucheng Xie Hao Gao Scene flow estimation from 3D point clouds based on dual‐branch implicit neural representations IET Computer Vision image enhancement image motion analysis image sensors learning (artificial intelligence) motion estimation object detection |
title | Scene flow estimation from 3D point clouds based on dual‐branch implicit neural representations |
title_full | Scene flow estimation from 3D point clouds based on dual‐branch implicit neural representations |
title_fullStr | Scene flow estimation from 3D point clouds based on dual‐branch implicit neural representations |
title_full_unstemmed | Scene flow estimation from 3D point clouds based on dual‐branch implicit neural representations |
title_short | Scene flow estimation from 3D point clouds based on dual‐branch implicit neural representations |
title_sort | scene flow estimation from 3d point clouds based on dual branch implicit neural representations |
topic | image enhancement image motion analysis image sensors learning (artificial intelligence) motion estimation object detection |
url | https://doi.org/10.1049/cvi2.12237 |
work_keys_str_mv | AT mingliangzhai sceneflowestimationfrom3dpointcloudsbasedondualbranchimplicitneuralrepresentations AT kangni sceneflowestimationfrom3dpointcloudsbasedondualbranchimplicitneuralrepresentations AT jiuchengxie sceneflowestimationfrom3dpointcloudsbasedondualbranchimplicitneuralrepresentations AT haogao sceneflowestimationfrom3dpointcloudsbasedondualbranchimplicitneuralrepresentations |