计算机科学

计算机科学 ›› 2025, Vol. 52 ›› Issue (7): 92-102.doi: 10.11896/jsjkx.250200011

• 计算机软件 • 上一篇    下一篇

基于运动模式与时频域融合的行人轨迹预测

刘亚俊, 纪庆革   

  1. 中山大学计算机学院 广州 510006
    广东省大数据分析与处理重点实验室 广州 510006
  • 收稿日期:2025-02-05 修回日期:2025-06-12 发布日期:2025-07-17
  • 通讯作者: 纪庆革(issjqg@mail.sysu.edu.cn)
  • 作者简介:(liuyj287@mail2.sysu.edu.cn)
  • 基金资助:
    国家自然科学基金(62276280);广东省自然科学基金(2016A030313288)

Pedestrian Trajectory Prediction Based on Motion Patterns and Time-Frequency Domain Fusion

LIU Yajun, JI Qingge   

  1. College of Computer Science and Engineering, Sun Yat-sen University, Guangzhou 510006, China
    Guangdong Key Laboratory of Big Data Analysis and Processing, Guangzhou 510006, China
  • Received:2025-02-05 Revised:2025-06-12 Published:2025-07-17
  • About author:LIU Yajun,born in 1990,postgraduate.His main research interests include computer vision and computer gra-phics.
    JI Qingge,born in 1966,Ph.D,associate professor,is a senior member of CCF(No.07014S).His main research interests include computer vision,computer graphics and virtual reality.
  • Supported by:
    National Natural Science Foundation of China(62276280) and Natural Science Foundation of Guangdong Pro-vince(2016A030313288).

PDF (PC)

22

摘要： 由于人类行为存在不确定性以及预测未来固有的多模态特点,如何区别预测轨迹的可能性与重要性成为不可避免的问题;行人运动模式则可以作为区分的基准特征。现有以运动模式作为切入点的研究非常匮乏;此外,以往研究局限于轨迹时域或频域单个维度,未能同时纳入研究。为此,提出了一种基于运动模式与时频域融合的行人轨迹预测模型MPTF,其由轨迹概率预测、回归预测与门控融合网络等组成。概率预测子网络提取轨迹时频域高维特征,结合运动模式以分类方式推理未来轨迹发生概率;回归预测子网络的时域处理分支挖掘行人社交关系,频域处理分支则着重关注不同频率分量对预测准确性的差异性影响;门控网络融合双维度推理特征,以回归方式预测多模态未来轨迹。公开数据集实验表明:MPTF在评估指标ADE与FDE上的总体性能达到研究前沿同等竞争力水平,在UNIV数据集上,ADE/FDE取得0.22/0.40的最优结果,在ETH数据集上,FDE提升8.5%,证明了结合时频域轨迹特征方法的有效性。

关键词: 行人轨迹预测, 运动模式, 频域特征, 注意力机制, Transformer

Abstract: Due to the uncertainty of human behaviour and the inherent multi-modality of predicting the future,it becomes an ine-vitable issue to discern the significance and likelihood of predicted pedestrian trajectories.Pedestrian motion patterns can be used as the benchmark characteristics for differentiation.In addition,most existing studies have examined pedestrian trajectories only in terms of the temporal dimension,ignoring the potentially great help of the frequency dimension of trajectories.This paper proposes a pedestrian trajectory prediction model based on the motion patterns and time-frequency domain fusion,called MPTF.The probabilistic prediction sub-network of MPTF extracts the time and frequency domain feature embedding of trajectories.By combining pedestrian motion patterns,it predicts the occurrence probability of future trajectories through a classification task.The time-domain branch of the regression prediction sub-network mines the social relationships among pedestrians,while the frequency-domain branch focuses on the contributions of different frequency components of the trajectories to the prediction.The gated fusion network fuses the features from these two dimensions and conducts regression inference to obtain multi-modal future tra-jectories.Experimental results on multiple public datasets show that the model has achieved a level equivalent to that of the latest studies in terms of the evaluation metrics of Average Displacement Error(ADE) and Final Displacement Error(FDE).On the UNIV dataset,the model obtains optimal results of 0.22 and 0.40 for ADE and FDE respectively.Moreover,on the ETH dataset,the FDE is improved by 8.5%.This validates the effectiveness of the method that combines time-frequency domain trajectory features.

Key words: Pedestrian trajectory prediction, Motion patterns, Frequency domain feature, Attention mechanism, Transformer

中图分类号: 

  • TP391
[1]ALAHI A,GOEL K,RAMANATHAN V,et al.Social LSTM:Human trajectory prediction in crowded spaces[C]//Procee-dings of the IEEE Conference on Computer Vision and Pattern Recognition(CVPR).IEEE,2016:961-971.
[2]GUPTA A,JOHNSON J,LI F F,et al.Social GAN:Socially Acceptable Trajectories with Generative Adversarial Networks[C]//2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition(ICCV).IEEE,2018:2255-2264.
[3]MOHAMED A,QIAN K,ELHOSEINY M,et al.Social-STGCNN:A Social Spatio-Temporal Graph Convolutional Neural Network forHuman Trajectory Prediction[C]//2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition(ICCV).IEEE,2020:14412-14420.
[4]ZHOU H,REN D,XIA H,et al.AST-GNN:An attention-based spatio-temporal graph neural network for Interaction-aware pedestrian trajectory prediction[J].Neurocomputing,2021,445:298-308.
[5]CHEN W,SANG H,WANG J,et al.STIGCN:spatial-temporal interaction-aware graph convolution network for pedestrian trajectory prediction[J].The Journal of Supercomputing,2024,80(8):10695-10719.
[6]HUANG Y,BI H,LI Z,et al.Stgat:Modelingspatial-temporal interactions for human trajectory prediction[C]//Proceedings of the IEEE/CVF International Conference on Computer Vision(ICCV).IEEE,2019:6272-6281.
[7]SALZMANN T,IVANOVIC B,CHAKRAVARTY P,et al.Trajectron++:Dynamically-feasible trajectory forecasting with heterogeneous data[C]//Computer Vision-ECCV 2020:16th European Conference.Springer,2020:683-700.
[8]SHI L,WANG L,LONG C,et al.Social interpretable tree for pedestrian trajectory prediction[C]//Proceedings of the AAAI Conference on Artificial Intelligence.2022:2235-2243.
[9]XU Y,PIAO Z,GAO S.Encoding crowd interaction with deep neural network for pedestrian trajectory prediction[C]//Proceedings of the IEEE Conference on Compu-ter Vision and Pattern Recognition(ICCV).IEEE,2018:5275-5284.
[10]YUE J,MANOCHA D,WANG H.Human trajectory prediction via neural social physics[C]//European Conference on Compu-ter Vision(ECCV).Springer,2022:376-394.
[11]PELLEGRINI S,ESS A,SCHINDLER K,et al.You'll neverwalk alone:Modeling social behaviorfor multi-target tracking[C]//2009 IEEE 12th International Conference on Computer Vision.IEEE,2009:261-268.
[12]LERNER A,CHRYSANTHOU Y,LISCHINSKI D.Crowds by example[C]//Computer Graphics Forum.Oxford,UK:Blackwell Publishing Ltd.,2007:655-664.
[13]ROBICQUET A,SADEGHIAN A,ALAHI A,et al.Learning social etiquette:Human trajectory understanding in crowded scenes[C]//Computer Vision-ECCV 2016:14th European Conference.Springer,2016:549-565.
[14]SUN J,LI Y X,FANG H S,et al.Three Steps to Multimodal Trajectory Prediction:Modality Clustering,Classification and Synthesis[C]//2021 IEEE/CVF International Conference on Computer Vision(ICCV).IEEE,2021:13230-13239.
[15]SHI L,WANG L,ZHOU S,et al.Trajectory Unified Transformer for Pedestrian Trajectory Prediction[C]//2023 IEEE/CVF International Conference on Computer Vision(ICCV).2023:9641-9650.
[16]YU C,MA X,REN J,et al.Spatio-temporalgraph transformer networks for pedestrian trajectory prediction[C]//Computer Vision-ECCV 2020:16th European Conference.Springer,2020:507-523.
[17]WONG C,XIA B,HONG Z,et al.View vertically:A hierarchical network for trajectory prediction via fourier spectrums[C]//European Conference on Computer Vision(ECCV).Springer,2022:682-700.
[18]LIU S,ZHU Y,YAO P,et al.SpectrumNet:Spectrum-BasedTrajectory Encode Neural Network for Pedestrian Trajectory Prediction[C]//ICASSP 2024－2024 IEEE International Confe-rence on Acoustics,Speech and Signal Processing(ICASSP).IEEE,2024:7075-7079.
[19]ZHOU T,MA Z,WEN Q,et al.Fedformer:Frequency enhanced decomposed transformer for long-term series forecasting[C]//International Conference on Machine Learning.PMLR,2022:27268-27286.
[20]ELDELE E,RAGAB M,CHEN Z,et al.Tslanet:Rethinkingtransformers for time series representation learning[J].arXiv:2404.08472,2024.
[21]LUO Y,ZHANG S,LYU Z,et al.Tfdnet:Time-frequency enhanced decomposed network for long-term time series forecasting[J].Pattern Recognition,2025,162:111412.
[22]YI K,ZHANG Q,FAN W,et al.Frequency-domain mlps are more effective learners in time series forecasting[J].Advances in Neural Information Processing Systems,2023,36:76656-76679.
[23]YUE W,LIU Y,YING X,et al.FreEformer:Frequency Enhanced Transformer for Multivariate Time Series Forecasting[J].arXiv:2501.13989,2025.
[24]SADEGHIAN A,KOSARAJU V,SADEGHIAN A,et al.So-phie:An attentive gan for predicting paths compliant to social and physical constraints[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition.2019:1349-1358.
[25]MANGALAM K,GIRASE H,AGARWAL S,et al.It is not the journey but the destination:Endpoint conditioned trajectory prediction[C]//Computer Vision-ECCV 2020:16th European Conference.Springer,2020:759-776.
[26]DUAN J,WANG L,LONG C,et al.Complementary attention gated network for pedestrian trajectory prediction[C]//Proceedings of the AAAI Conference on Artificial Intelligence.2022:542-550.
[27]XU C,MAO W,ZHANG W,et al.Remember intentions:Retrospective-memory-based trajectory prediction[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition.IEEE,2022:6488-6497.
[28]XU C,LI M,NI Z,et al.Groupnet:Multiscale hypergraph neuralnetworks for trajectory prediction with relational reasoning[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition.2022:6498-6507.
[29]XU C,TAN R T,TAN Y,et al.Eqmotion:Equivariant multi-agent motion prediction with invariant interaction reasoning[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition(ICCV).IEEE,2023:1410-1420.
[30]DONG Y,WANG L,ZHOU S,et al.Recurrent Aligned Network for Generalized Pedestrian Trajectory Prediction[J].ar-Xiv:2403.05810,2024.
[31]WANG M,ZHU X,YU C,et al.Ganet:Goal area network for motion forecasting[C]//2023 IEEE International Conference on Robotics and Automation(ICRA).IEEE,2023:1609-1615.
[32]MANGALAM K,AN Y,GIRASE H,et al.From goals,way-points & paths to long termhuman trajectory forecasting[C]//Proceedings of the IEEE/CVF International Conference on Computer Vision.2021:15233-15242.
[33]CAO Z,GAO H,MANGALAM K,et al.Long-term human motion prediction with scenecontext[C]//Computer Vision-ECCV 2020:16th European Conference.Springer,2020:387-404.
[34]XU P,HAYET J B,KARAMOUZAS I.Socialvae:Human trajectory prediction using timewise latents[C]//European Confe-rence on Computer Vision.Cham:Springer,2022:511-528.
[35]LIANG J,JIANG L,NIEBLES J C,et al.Peeking into the future:Predicting future person activities and locations in videos[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition.2019:5725-5734.
Viewed
Full text


Abstract

Cited
  Shared   
  Discussed   
No Suggested Reading articles found!
渝ICP备16013121号-4
地址：重庆市渝北区洪湖西路18号    邮编：401121  
电话：023-63500828（编辑部） 023-67039612（会议/专题） 023-67039623（财务/发票）
E-mail：jsjkx12@163.com
本系统由北京玛格泰克科技发展有限公司设计开发