RECO: Rotation Equivariant COnvolutional Neural Network for Human Trajectory Forecasting

Jijun Cheng; Hao Wang; Dongheng Shao; Jian Yang; Mingsong Chen; Xian Wei; Xuan Tang

doi:10.1007/978-981-99-8435-0_39

RECO: Rotation Equivariant COnvolutional Neural Network for Human Trajectory Forecasting

Jijun Cheng, Hao Wang, Dongheng Shao, Jian Yang, Mingsong Chen, Xian Wei, Xuan Tang^*

^*Corresponding author for this work

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

Abstract

Pedestrian trajectory prediction is a crucial task for various applications, especially for autonomous vehicles that need to ensure safe navigation. To perform this task, it is essential to understand the dynamics of pedestrian motion and account for the uncertainty and multimodality of human behaviors. However, existing methods often produce inconsistent and unrealistic predictions due to their limited ability to handle different orientations of pedestrians. In this paper, we propose a novel approach that leverages the Euclidean group C4 to enhance convolutional neural networks with rotation equivariance. This property enables the networks to learn features that are invariant to rotations, thus maintaining consistent output under various orientations. We present the Rotation Equivariant COnvolutional Neural Network (RECO), a model specifically designed for pedestrian trajectory prediction using rotation equivariant convolutions. We evaluate our model on challenging real-world human trajectory forecasting datasets and show that it achieves competitive performance compared to state-of-the-art methods.

Original language	English
Title of host publication	Pattern Recognition and Computer Vision - 6th Chinese Conference, PRCV 2023, Proceedings
Editors	Qingshan Liu, Hanzi Wang, Rongrong Ji, Zhanyu Ma, Weishi Zheng, Hongbin Zha, Xilin Chen, Liang Wang
Publisher	Springer Science and Business Media Deutschland GmbH
Pages	492-504
Number of pages	13
ISBN (Print)	9789819984343
DOIs	https://doi.org/10.1007/978-981-99-8435-0_39
State	Published - 2024
Event	6th Chinese Conference on Pattern Recognition and Computer Vision, PRCV 2023 - Xiamen, China Duration: 13 Oct 2023 → 15 Oct 2023

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	14427 LNCS
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Conference

Conference	6th Chinese Conference on Pattern Recognition and Computer Vision, PRCV 2023
Country/Territory	China
City	Xiamen
Period	13/10/23 → 15/10/23

Keywords

Human Trajectory Forecasting
Multimodality
Rotation Equivariant Convolution
Uncertainty

Access to Document

10.1007/978-981-99-8435-0_39

Cite this

Cheng, J., Wang, H., Shao, D., Yang, J., Chen, M., Wei, X., & Tang, X. (2024). RECO: Rotation Equivariant COnvolutional Neural Network for Human Trajectory Forecasting. In Q. Liu, H. Wang, R. Ji, Z. Ma, W. Zheng, H. Zha, X. Chen, & L. Wang (Eds.), Pattern Recognition and Computer Vision - 6th Chinese Conference, PRCV 2023, Proceedings (pp. 492-504). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 14427 LNCS). Springer Science and Business Media Deutschland GmbH. https://doi.org/10.1007/978-981-99-8435-0_39

Cheng, Jijun ; Wang, Hao ; Shao, Dongheng et al. / RECO : Rotation Equivariant COnvolutional Neural Network for Human Trajectory Forecasting. Pattern Recognition and Computer Vision - 6th Chinese Conference, PRCV 2023, Proceedings. editor / Qingshan Liu ; Hanzi Wang ; Rongrong Ji ; Zhanyu Ma ; Weishi Zheng ; Hongbin Zha ; Xilin Chen ; Liang Wang. Springer Science and Business Media Deutschland GmbH, 2024. pp. 492-504 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inproceedings{dacd64549eb64651aa04450ac2c44e8f,

title = "RECO: Rotation Equivariant COnvolutional Neural Network for Human Trajectory Forecasting",

abstract = "Pedestrian trajectory prediction is a crucial task for various applications, especially for autonomous vehicles that need to ensure safe navigation. To perform this task, it is essential to understand the dynamics of pedestrian motion and account for the uncertainty and multimodality of human behaviors. However, existing methods often produce inconsistent and unrealistic predictions due to their limited ability to handle different orientations of pedestrians. In this paper, we propose a novel approach that leverages the Euclidean group C4 to enhance convolutional neural networks with rotation equivariance. This property enables the networks to learn features that are invariant to rotations, thus maintaining consistent output under various orientations. We present the Rotation Equivariant COnvolutional Neural Network (RECO), a model specifically designed for pedestrian trajectory prediction using rotation equivariant convolutions. We evaluate our model on challenging real-world human trajectory forecasting datasets and show that it achieves competitive performance compared to state-of-the-art methods.",

keywords = "Human Trajectory Forecasting, Multimodality, Rotation Equivariant Convolution, Uncertainty",

author = "Jijun Cheng and Hao Wang and Dongheng Shao and Jian Yang and Mingsong Chen and Xian Wei and Xuan Tang",

note = "Publisher Copyright: {\textcopyright} 2024, The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd.; 6th Chinese Conference on Pattern Recognition and Computer Vision, PRCV 2023 ; Conference date: 13-10-2023 Through 15-10-2023",

year = "2024",

doi = "10.1007/978-981-99-8435-0\_39",

language = "英语",

isbn = "9789819984343",

series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

publisher = "Springer Science and Business Media Deutschland GmbH",

pages = "492--504",

editor = "Qingshan Liu and Hanzi Wang and Rongrong Ji and Zhanyu Ma and Weishi Zheng and Hongbin Zha and Xilin Chen and Liang Wang",

booktitle = "Pattern Recognition and Computer Vision - 6th Chinese Conference, PRCV 2023, Proceedings",

address = "德国",

}

Cheng, J, Wang, H, Shao, D, Yang, J, Chen, M , Wei, X & Tang, X 2024, RECO: Rotation Equivariant COnvolutional Neural Network for Human Trajectory Forecasting. in Q Liu, H Wang, R Ji, Z Ma, W Zheng, H Zha, X Chen & L Wang (eds), Pattern Recognition and Computer Vision - 6th Chinese Conference, PRCV 2023, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 14427 LNCS, Springer Science and Business Media Deutschland GmbH, pp. 492-504, 6th Chinese Conference on Pattern Recognition and Computer Vision, PRCV 2023, Xiamen, China, 13/10/23. https://doi.org/10.1007/978-981-99-8435-0_39

RECO: Rotation Equivariant COnvolutional Neural Network for Human Trajectory Forecasting. / Cheng, Jijun; Wang, Hao; Shao, Dongheng et al.
Pattern Recognition and Computer Vision - 6th Chinese Conference, PRCV 2023, Proceedings. ed. / Qingshan Liu; Hanzi Wang; Rongrong Ji; Zhanyu Ma; Weishi Zheng; Hongbin Zha; Xilin Chen; Liang Wang. Springer Science and Business Media Deutschland GmbH, 2024. p. 492-504 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 14427 LNCS).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - RECO

T2 - 6th Chinese Conference on Pattern Recognition and Computer Vision, PRCV 2023

AU - Cheng, Jijun

AU - Wang, Hao

AU - Shao, Dongheng

AU - Yang, Jian

AU - Chen, Mingsong

AU - Wei, Xian

AU - Tang, Xuan

PY - 2024

Y1 - 2024

N2 - Pedestrian trajectory prediction is a crucial task for various applications, especially for autonomous vehicles that need to ensure safe navigation. To perform this task, it is essential to understand the dynamics of pedestrian motion and account for the uncertainty and multimodality of human behaviors. However, existing methods often produce inconsistent and unrealistic predictions due to their limited ability to handle different orientations of pedestrians. In this paper, we propose a novel approach that leverages the Euclidean group C4 to enhance convolutional neural networks with rotation equivariance. This property enables the networks to learn features that are invariant to rotations, thus maintaining consistent output under various orientations. We present the Rotation Equivariant COnvolutional Neural Network (RECO), a model specifically designed for pedestrian trajectory prediction using rotation equivariant convolutions. We evaluate our model on challenging real-world human trajectory forecasting datasets and show that it achieves competitive performance compared to state-of-the-art methods.

AB - Pedestrian trajectory prediction is a crucial task for various applications, especially for autonomous vehicles that need to ensure safe navigation. To perform this task, it is essential to understand the dynamics of pedestrian motion and account for the uncertainty and multimodality of human behaviors. However, existing methods often produce inconsistent and unrealistic predictions due to their limited ability to handle different orientations of pedestrians. In this paper, we propose a novel approach that leverages the Euclidean group C4 to enhance convolutional neural networks with rotation equivariance. This property enables the networks to learn features that are invariant to rotations, thus maintaining consistent output under various orientations. We present the Rotation Equivariant COnvolutional Neural Network (RECO), a model specifically designed for pedestrian trajectory prediction using rotation equivariant convolutions. We evaluate our model on challenging real-world human trajectory forecasting datasets and show that it achieves competitive performance compared to state-of-the-art methods.

KW - Human Trajectory Forecasting

KW - Multimodality

KW - Rotation Equivariant Convolution

KW - Uncertainty

UR - https://www.scopus.com/pages/publications/85180745524

U2 - 10.1007/978-981-99-8435-0_39

DO - 10.1007/978-981-99-8435-0_39

M3 - 会议稿件

AN - SCOPUS:85180745524

SN - 9789819984343

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 492

EP - 504

BT - Pattern Recognition and Computer Vision - 6th Chinese Conference, PRCV 2023, Proceedings

A2 - Liu, Qingshan

A2 - Wang, Hanzi

A2 - Ji, Rongrong

A2 - Ma, Zhanyu

A2 - Zheng, Weishi

A2 - Zha, Hongbin

A2 - Chen, Xilin

A2 - Wang, Liang

PB - Springer Science and Business Media Deutschland GmbH

Y2 - 13 October 2023 through 15 October 2023

ER -

Cheng J, Wang H, Shao D, Yang J, Chen M , Wei X et al. RECO: Rotation Equivariant COnvolutional Neural Network for Human Trajectory Forecasting. In Liu Q, Wang H, Ji R, Ma Z, Zheng W, Zha H, Chen X, Wang L, editors, Pattern Recognition and Computer Vision - 6th Chinese Conference, PRCV 2023, Proceedings. Springer Science and Business Media Deutschland GmbH. 2024. p. 492-504. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). doi: 10.1007/978-981-99-8435-0_39