计算机科学

计算机科学 ›› 2025, Vol. 52 ›› Issue (12): 133-140.doi: 10.11896/jsjkx.241200212

• 计算机图形学&多媒体 • 上一篇    下一篇

构建场景-行人-行人交互的行人轨迹预测时空图卷积网络

洪铭骏, 纪庆革   

  1. 中山大学计算机学院 广州 510006
  • 收稿日期:2024-12-30 修回日期:2025-05-09 出版日期:2025-12-15 发布日期:2025-12-09
  • 通讯作者: 纪庆革(issjqg@mail.sysu.edu.cn)
  • 作者简介:(hongmj6@mail2.sysu.edu.cn)
  • 基金资助:
    广东省自然科学基金(2016A030313288)

SPP-STGCN:Spatio-Temporal Graph Convolutional Network for Pedestrian Trajectory Predictionwith Scene-Perdestrian-Perdestrain Interactions

HONG Mingjun, JI Qingge   

  1. School of Computer Science and Engineering, Sun Yat-sen University, Guangzhou 510006, China
  • Received:2024-12-30 Revised:2025-05-09 Published:2025-12-15 Online:2025-12-09
  • About author:HONG Mingjun,born in 2003,postgraduate.His main research interests include deep learning and computer vision.
    JI Qingge,born in 1966,Ph.D,associate professor,is a senior member of CCF(No.07014S).His main research in-terests include computer vision,computer graphics and virtual reality.
  • Supported by:
    This work was supported by the Natural Science Foundation of Guangdong Province,China(2016A030313288).

PDF (PC)

11

摘要： 行人轨迹预测是自动驾驶和智能监控系统中的一项基础而关键的任务。场景的限制是影响行人运动轨迹的重要因素之一。尽管现有的研究已经尝试将场景因素融入轨迹预测中,但这些方法在整合场景信息时往往存在不足,尤其是没有考虑到场景的全面融合。对此,提出了一种新的行人轨迹预测模型——构建场景-行人-行人交互的时空图卷积网络SPP-STGCN(Spatio-Temporal Graph Convolutional Network with Scene-Perdestrian-Perdestrain Interactions)。SPP-STGCN模型采用两阶段架构来提高预测准确性。第一阶段,模型将轨迹和场景数据整合输入,通过场景邻接矩阵融合块SAFB实现两种特征的融合,构建出融合场景特征的时空图邻接矩阵,为预测提供丰富的上下文信息。同时,模型在时间、空间两个维度并行执行,结合轨迹信息分别构造出描述时间相关性的和空间相关性的行人轨迹时空图。第二阶段,场景图卷积网络对时间与空间维度的时空图进行特征提取。提取的特征随后被融合,并通过时间金字塔外推卷积进行处理,以获得行人未来轨迹的二维高斯分布。最后SPP-STGCN以该分布作为预测行人轨迹的概率模型,通过采样生成行人未来轨迹。在ETH和UCY公开数据集上的对比实验结果显示,SPP-STGCN模型在与9种主流模型的对比实验中的表现达到了目前的最佳水平。消融实验与定性分析进一步证实了所提模型的有效性与合理性。SPP-STGCN模型通过有效整合场景特征,显著提升了行人轨迹预测的性能。

关键词: 行人轨迹预测, 场景-行人-行人交互, 时空图卷积, 邻接矩阵生成, 注意力机制

Abstract: Pedestrian trajectory prediction is a fundamental and critical task in autonomous driving and intelligent surveillance systems.The constraints of the scene are one of the important factors affecting pedestrian movement trajectories.Despite existing research efforts to incorporate scene factors into trajectory prediction,these methods often fall short in integrating scene information,particularly in terms of comprehensive scene fusion.To overcome these limitations,this study proposes a new pedestrian trajectory prediction model,namely SPP-STGCN.The SPP-STGCN model adopts a two-stage architecture to enhance prediction accuracy.In the first stage,the model integrates trajectory and scene data.Through the Scene Adjacency Fusion Block(SAFB),the model fuses these two types of features to construct a spatio-temporal graph adjacency matrix that incorporates scene features,thereby providing rich contextual information for prediction.Concurrently,the model operates in parallel along the temporal and spatial dimensions,constructing pedestrian trajectory spatio-temporal graphs that describe temporal and spatial correlations based on trajectory information.In the second stage,scene-graph convolutional networks extract features from the temporal and spatial spatio-temporal graphs.The extracted features are then fused and processed through a temporal pyramid extrapolation convolution to obtain the two-dimensional Gaussian distribution of the pedestrian’s future trajectory.Finally,SPP-STGCN uses this distribution as a probabilistic model for predicting pedestrian trajectories,generating future trajectories through sampling.Comparative experimental results on the ETH and UCY public datasets show that the SPP-STGCN model has achieved the current state-of-the-art performance in comparison experiments with nine mainstream models.Ablation experiments and qualitative analysis further confirm the effectiveness and rationality of the proposed model.The SPP-STGCN model significantly enhances pedestrian trajectory prediction performance by effectively integrating scene features.

Key words: Pedestrian trajectory prediction, Scene-pedestrian-pedestrian interaction, Spatio-temporal graph convolution, Adjacency matrix generation, Attention mechanism

中图分类号: 

  • TP391
[1]HELBING D,MOLNAR P.Social Force Model for Pedestrian Dynamics[J].Physical Review E,1995,51(5):4282-4286.
[2]ALAHI A,GOEL K,RAMANATHAN V,et al.Social LSTM:human trajectory prediction in crowded spaces[C]//Proceedings of 2016 IEEE Conference on Computer Vision and Pattern Re-cognition.IEEE,2016:961-971.
[3]MOHAMED A,QIAN K,ELHOSEINY M,et al.Social-STGCNN:A Social Spatio-Temporal Graph Convolutional Neural Network for Human Trajectory Prediction[C]//Proceedings of 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition.IEEE,2020:14412-14420.
[4]SHI L,WANG L,LONG C,et al.SGCN:Sparse Graph Convolution Network for Pedestrian Trajectory Prediction[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition.IEEE,2021:8994-9003.
[5]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 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition.IEEE,2019:1349-1358.
[6]MANGALAM K,AN Y,GIRASE H,et al.From Goals,Waypoints & Paths to Long Term Human Trajectory Forecasting[C]//Proceedings of 2021 IEEE/CVF International Conference on Computer Vision.IEEE,2021:15213-15222.
[7]CHEN H,GI Q.Scene-constrained spatial-temporal graph con-volutional network for pedestrian trajectory prediction[J].Journal of Image and Graphics,2023,28(10):3163-3175.
[8]ZHANG P,OUYANG W,ZHANG P,et al.SR-LSTM:StateRefinement for LSTM Towards Pedestrian Trajectory Prediction[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition.IEEE,2019:12085-12094.
[9]LIAN J,REN W,LI L,et al.PTP-STGCN:Pedestrian Trajectory Prediction Based on a Spatio-temporal Graph Convolutional Neural Network[J].Applied Intelligence,2023,53:2862-2878.
[10]KOSARAJU V,SADEGHIAN A,MARTÍN-MARTÍN R,et al.Social-BiGAT:multimodal trajectory forecasting using Bicycle-GAN and graph attention networks[J].arXiv:1907.03395,2019.
[11]KIPF T N,WELLING M.Semi-Supervised Classification withGraph Convolutional Networks [J].arXiv:1609.02907,2016.
[12]CHEN W,SANG H,WANG J,et al.STIGCN:spatial-temporal interaction-aware graph convolution network for pedestrian tra-jectory prediction[J].The Journal of Supercomputing,2024,80(8):10695-10719.
[13]HUANG R,XUE H,PAGNUCCO M,et al.Multimodal Trajectory Prediction:A Survey[J].arXiv:2302.10463,2023.
[14]GUPTA A,JOHNSON J,FEI-FEI L,et al.Social GAN:Socially Acceptable Trajectories with Generative Adversarial Networks[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition.2018:2255-2264.
[15]GOODFELLOW I,POUGET-ABADIE J,MIRZA M,et al.Ge-nerative Adversarial Nets[EB/OL].https://proceedings.neurips.cc/paper_files/paper/2014/file/5ca3e9b122f61f8f06494c97b1afccf3-Paper.pdf.
[16]YUE J,MANOCHA D,WANG H.Human Trajectory Prediction via Neural Social Physics[C]//European Conference on Computer Vision.Cham:Springer:2022:376-394.
[17]XU P,HAYET J B,KARAMOUZAS I.SocialVAE:HumanTrajectory Prediction using Timewise Latents[C]//Proceedings of European Conference on Computer Vision.Cham:Springer,2022:511-528.
[18]ZHOU H,YANG X,REN D,et al.CSIR:Cascaded SlidingCVAEs With Iterative Socially-Aware Rethinking for Trajectory Prediction[J].IEEE Transactions on Intelligent Transportation Systems,2023(12):24.
[19]XIANG W,YIN H,WANG H,et al.SocialCVAE:Predicting Pedestrian Trajectory via Interaction Conditioned Latents[C]//Proceedings of the AAAI Conference on Artificial Intelligence.2024:6216-6224.
[20]MANH H,ALAGHBAND G.Scene-LSTM:A Model for Human Trajectory Prediction.[J].arXiv:1808.04018,2018.
[21]VASWANI A,SHAZEER N,PARMAR N,et al.Attention Is All You Need[C]//Proceedings of the 31st International Conference on Neural Information Processing Systems.2017:6000-6010.
[22]KIM T,KIM J,TAE Y,et al.Reversible Instance Normalization for Accurate Time-Series Forecasting against Distribution Shift[EB/OL].https://openreview.net/pdf?id=cGDAkQo1C0p.
[23]PELLEGRINI S,ESS A,GOOL L V.Improving Data Association by Joint Modeling of Pedestrian Trajectories and Groupings[C]//Proceedings of the 11th European Conference on Compu-ter Vision.Springer,2010:452-465.
[24]LERNER A,CHRYSANTHOU Y,LISCHINSKI D.Crowds by Example[J].Computer Graphics Forum,2007,26(3):655-664.
[25]HUANG Y,BI H,LI Z,et al.STGAT:Modeling Spatial-Temporal Interactions for Human Trajectory Prediction[C]//Proceedings of the IEEE/CVF International Conference on Computer Vision.2019:6272-6281.
[26]YU C,MA X,REN J,et al.Spatio-Temporal Graph Transfor-mer Networks for Pedestrian Trajectory Prediction[C]//Procee-dings of Computer Vision-ECCV:16th European Conference.Cham:Springer,2020:507-523.
[27]CHEN G,LI J,LU J,et al.Human Trajectory Prediction viaCounterfactual Analysis[C]//Proceedings of the IEEE/CVF International Conference on Computer Vision.2021:9824-9833.
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