The DOI matches its Crossref record; 23 of the 25 references it cites were found in a registry or an index.
You'll never walk alone: Modeling social behavior for multi-target tracking
Abstract
Object tracking typically relies on a dynamic model to predict the object's location from its past trajectory. In crowded scenarios a strong dynamic model is particularly important, because more accurate predictions allow for smaller search regions, which greatly simplifies data association. Traditional dynamic models predict the location for each target solely based on its own history, without taking into account the remaining scene objects. Collisions are resolved only when they happen. Such an approach ignores important aspects of human behavior: people are driven by their future destination, take into account their environment, anticipate collisions, and adjust their trajectories at an early stage in order to avoid them. In this work, we introduce a model of dynamic social behavior, inspired by models developed for crowd simulation. The model is trained with videos recorded from birds-eye view at busy locations, and applied as a motion model for multi-people tracking from a vehicle-mounted camera. Experiments on real sequences show that accounting for social interactions and scene knowledge improves tracking performance, especially during occlusions.
Overview
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
Citations
Cites
Histograms of Oriented Gradients for Human Detection
Dalal, N., & Triggs, B. (2005) · 2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR'05) · doi:10.1109/cvpr.2005.177 · Cited by 33,584
Social force model for pedestrian dynamics.
Helbing, D., & Molnár, P. (1995) · Physical review. E, Statistical physics, plasmas, fluids, and related interdisciplinary topics · doi:10.1103/physreve.51.4282 · Cited by 7,299
A discriminatively trained, multiscale, deformable part model
Felzenszwalb, P. F., McAllester, D., & Ramanan, D. (2008) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr.2008.4587597 · Cited by 2,972
Evaluating Multiple Object Tracking Performance: The CLEAR MOT Metrics
Bernardin, K., & Stiefelhagen, R. (2008) · EURASIP Journal on Image and Video Processing · doi:10.1155/2008/246309 · Cited by 2,918
Abnormal crowd behavior detection using social force model
Mehran, R., Oyama, A., & Shah, M. A. (2009) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr.2009.5206641 · Cited by 1,698
Lerner, A., Chrysanthou, Y., & Lischinski, D. (2007) · Computer Graphics Forum · doi:10.1111/j.1467-8659.2007.01089.x · Cited by 1,199
A Boosted Particle Filter: Multitarget Detection and Tracking
Okuma, K., Taleghani, A., de Freitas, N., Little, J. J., & Lowe, D. G. (2004) · Lecture notes in computer science · doi:10.1007/978-3-540-24670-1_3 · Cited by 1,025
Gräbner, H., & Bischof, H. (2006) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr.2006.215 · Cited by 998
Global data association for multi-object tracking using network flows
Zhang, L., Li, Y., & Nevatia, R. (2008) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr.2008.4587584 · Cited by 945
People-tracking-by-detection and people-detection-by-tracking
Andriluka, M., Roth, S., & Schiele, B. (2008) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr.2008.4587583 · Cited by 825
Flagged first, then by citations
2 of the 25 references couldn’t be matched to a known paper.
Research works printed but not linked: 1
- schadschneider, 2001, Cellular automaton approach to pedestrian dynamics—theory, PED
Cited by
Sorted by citations
Items 401–500 of 1,000
LEMMA: A Multi-view Dataset for L Earning Multi-agent Multi-task Activities
Jia, B., Chen, Y., Huang, S., Zhu, Y., & Zhu, S. C. (2020) · Lecture notes in computer science · doi:10.1007/978-3-030-58574-7_46 · Cited by 36
“Another related stream of work is the learning of group-level activities in a multi-agent setting [26], such as detecting key actors [43], predicting future trajectories [40,34], and recognizing collective activities [11,39,50].”
BackgroundCounting People by Estimating People Flows
Liu, W., Salzmann, M., & Fua, P. (2021) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2021.3102690 · Cited by 39
“Imposing strong conservation constraints when tracking people has been a concern long before the advent of deep learning [56], [57], [58], [59], [60], [61], [62], [63], [64], [65], [66], [67].”
Antonucci, A., Papini, G. P. R., Bevilacqua, P., et al. (2021) · IEEE Access · doi:10.1109/access.2021.3138614 · Cited by 35
“To further substantiate the analysis, we experimentally validate the performance of theNet1 network andmake a comparison with other methods in the literature, we used two widely known human motion datasets: the ETH [37] dataset (with the scene Hotel and ETH) and UCY [38] dataset (with scene UCY, Zara1 and Zara2).”
MethodDeep Learning for Vision-based Prediction: A SurveyPreprint
Rasouli, A. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2007.00095 · Cited by 51
“For instance, in the context of surveillance, ETH [235] is one of the most commonly used datasets.”
BackgroundInfluentialGenerating Critical Test Scenarios for Autonomous Driving Systems via Influential Behavior Patterns
Tian, H., Wu, G., Yan, J., et al. (2022) · IEEE/ACM International Conference on Automated Software Engineering (ASE) · doi:10.1145/3551349.3560430 · Cited by 34
“We investigated multiple public larg -scale traffic trajectory datasets [10, 21, 25, 27, 38, 38, 39, 48], and chose Stanford Campus [39] and inD [10], as these two datasets contain far more trajectories than others.”
BackgroundChen, W., Sang, H., Wang, J., et al. (2024) · Information Sciences · doi:10.1016/j.ins.2024.120455 · Cited by 34
Liu, Y., Dong, X., Lin, Y., Ye, M., Zhang, K., & Du, B. (2025) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2025.3645918 · Cited by 43
“Two commonly reported metrics for evaluating prediction performance are Average Displacement Error (ADE) and Final Displacement Error (FDE) [60].”
Leveraging Long-Term Predictions and Online Learning in Agent-Based Multiple Person Tracking
Liu, W., Chan, A. B., Lau, R. W. H., et al. (2014) · IEEE Transactions on Circuits and Systems for Video Technology · doi:10.1109/tcsvt.2014.2344511 · Cited by 34
“We use the same crowd datasets used in [30] and [38] to study real-world interactions among pedestrians and evaluate motion models.”
BackgroundMethodInfluentialAutomatic model construction for the behavior of human crowds
Zhong, J., Cai, W., Lees, M., et al. (2017) · Applied Soft Computing · doi:10.1016/j.asoc.2017.03.020 · Cited by 34
Social graph convolutional LSTM for pedestrian trajectory prediction
Zhou, Y., Wu, H., Cheng, H., Qi, K., Hu, K., Kang, C., & Zheng, J. (2021) · IET Intelligent Transport Systems · doi:10.1049/itr2.12033 · Cited by 34
“The graph convolution network [24, 25] is a powerful neural network architecture for machine learning on graphs.”
BackgroundAn improved GAN with transformers for pedestrian trajectory prediction models
Lv, Z., Huang, X., & Cao, W. (2021) · International Journal of Intelligent Systems · doi:10.1002/int.22724 · Cited by 34
“The input sequence is 3.2 s (8 steps) in total, while the predicted sequence is the next 4.8 s (12 steps), which is similar to the setting of References [5,46].”
MethodInfluentialRobots That Can See: Leveraging Human Pose for Trajectory Prediction
Salzmann, T., Chiang, H.-T. L., Ryll, M., Sadigh, D., Parada, C., & Bewley, A. (2023) · IEEE Robotics and Automation Letters · doi:10.1109/lra.2023.3312035 · Cited by 33
“In addition, we also compare our model to the ETH [5] and UCY [6] datasets.”
BackgroundMethodResultInfluentialWTGCN: wavelet transform graph convolution network for pedestrian trajectory prediction
Chen, W., Sang, H., Wang, J., et al. (2024) · International Journal of Machine Learning and Cybernetics · doi:10.1007/s13042-024-02258-5 · Cited by 33
Progressive Pretext Task Learning for Human Trajectory Prediction
Lin, X., Liang, T., Lai, J., & Hu, J. (2024) · Lecture notes in computer science · doi:10.1007/978-3-031-73404-5_12 · Cited by 33
Zitouni, M. S., Sluzek, A. S., & Bhaskar, H. (2019) · Engineering Applications of Artificial Intelligence · doi:10.1016/j.engappai.2019.04.012 · Cited by 32
Multiple Object Tracking with Motion and Appearance Cues
Li, W., Mu, J., & Liu, G. (2019) · IEEE International Conference on Computer Vision Workshops · doi:10.1109/iccvw.2019.00025 · Cited by 36
“In earlier works, the color histograms [18, 19] and some hand-crafted features [20, 21] are commonly used as descriptors of the appearance of objects.”
MethodChen, K., Song, X., & Ren, X. (2020) · IEEE Transactions on Circuits and Systems for Video Technology · doi:10.1109/tcsvt.2020.3013254 · Cited by 33
“The errors are measured in the pixel space on MOT whereas in meters on ETH [22] and UCY [5] because MOT dataset does not provide the homograph matrix [13].”
BackgroundMethodInfluentialDifferential evolution with sensitivity analysis and the Powell's method for crowd model calibration
Zhong, J., & Cai, W. (2015) · Journal of Computational Science · doi:10.1016/j.jocs.2015.04.013 · Cited by 33
“[5] proposed a modified social-force model to simulate pedestrian behaviors, so as to improve the tracking performance.”
BackgroundMethodMultiple Object Tracking by Flowing and FusingPreprint
Zhang, J., Zhou, S., Chang, X., Wan, F., Wang, J., Wu, Y., & Huang, D. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2001.11180 · Cited by 33
“In the earlier tracking approaches, the color histogram [12, 36, 33] and pixel-based template representations [64, 46] are standard hand-crafted features used to describe the appearance of moving objects.”
MethodLv, K., Yuan, L., & Ni, X. (2024) · IEEE Transactions on Instrumentation and Measurement · doi:10.1109/tim.2024.3375973 · Cited by 35
“3) Experimental results, in comparison to the state-of-the-art methods, demonstrate the superior performance of our proposed method on public datasets (ETH [34] and UCY [35]).”
MethodResultAttention-Aware Social Graph Transformer Networks for Stochastic Trajectory Prediction
Liu, Y., Li, B., Wang, X., Sammut, C., & Yao, L. (2024) · IEEE Transactions on Knowledge and Data Engineering · doi:10.1109/tkde.2024.3390765 · Cited by 33
“The benchmark datasets utilized in our study are ETH [39] and UCY [40], encompassing videos capturing pedestrian trajectories in real-world scenarios characterized by high social interactions.”
MethodMART: MultiscAle Relational Transformer Networks for Multi-agent Trajectory Prediction
Lee, S., Lee, J., Yu, Y., et al. (2024) · Lecture notes in computer science · doi:10.1007/978-3-031-72848-8_6 · Cited by 32
Vision-Based Multi-Future Trajectory Prediction: A Survey
Huang, R., Xue, H., Pagnucco, M., Salim, F. D., & Song, Y. (2025) · IEEE Transactions on Neural Networks and Learning Systems · doi:10.1109/tnnls.2025.3550350 · Cited by 64
“Pedestrian trajectory prediction studies mostly use ETH&UCY [32, 33] and Stanford Drone Dataset (SDD) [31] while vehicle trajectory prediction often uses InD [120], NuScenes [29], Argov-erse [28], INTERACTION [121] and Waymo [27] datasets.”
Discovering social interactions in real work environments
Chen, C., Cilla, R., Wu, C., et al. (2011) · doi:10.1109/fg.2011.5771376 · Cited by 32
“Researchers have also shown that understanding social interactions can be a basis for improving the performance of many existing systems, such as video tracking (by modeling the dynamics of the crowd [8]) and face recognition (by learning the interpersonal relationship of the people in the pictures [11]).”
BackgroundOnline multi‐person tracking with two‐stage data association and online appearance model learning
Ju, J., Kim, D., Ku, B., Han, D. K., & Ko, H. (2016) · IET Computer Vision · doi:10.1049/iet-cvi.2016.0068 · Cited by 32
Survey on Vision-Based Path Prediction
Hirakawa, T., Yamashita, T., Tamaki, T., & Fujiyoshi, H. (2018) · Lecture notes in computer science · doi:10.1007/978-3-319-91131-1_4 · Cited by 33
“The UCY Dataset [40] and ETH Dataset [41] contain videos of pedestrians walking along streets where no other moving objects exist, which is a relatively simple situation compared with situations of other datasets.”
BackgroundWinding Through: Crowd Navigation via Topological Invariance
Mavrogiannis, C., Balasubramanian, K., Poddar, S., et al. (2022) · IEEE Robotics and Automation Letters · doi:10.1109/lra.2022.3223024 · Cited by 32
“from datasets [20], [30] and crowd simulation engines [3], [43].”
BackgroundFrom Crowd Motion Prediction to Robot Navigation in Crowds
Poddar, S., Mavrogiannis, C., & Srinivasa, S. S. (2023) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros55552.2023.10341464 · Cited by 32
“2: Error in trajectory prediction of humans on the ETH [29] and UCY [20] datasets.”
BackgroundMethodInfluentialModeling local behavior for predicting social interactions towards human tracking
Yan, X., Kakadiaris, I. A., & Shah, S. K. (2013) · Pattern Recognition · doi:10.1016/j.patcog.2013.10.019 · Cited by 32
“The LTA model incorporates the human dynamics attributed to the behavior where a pedestrian tries to avoid running into other pedestrians while approaching ones destination.”
BackgroundMethodMRGTraj: A Novel Non-Autoregressive Approach for Human Trajectory Prediction
Peng, Y., Zhang, G., Shi, J., Li, X., & Zheng, L. (2023) · IEEE Transactions on Circuits and Systems for Video Technology · doi:10.1109/tcsvt.2023.3307442 · Cited by 33
“The model is trained and tested in ETH [54] and UCY [55] datasets, which consist of five scenarios (ETH, HOTEL, UNIV, ZARA1, and ZARA2).”
MethodPedestrian Trajectory Prediction Based on Social Interactions Learning With Random Weights
Xie, J., Zhang, S., Xia, B., Xiao, Z. Q., Jiang, H., Zhou, S., Qin, Z., & Chen, H. (2024) · IEEE Transactions on Multimedia · doi:10.1109/tmm.2024.3368931 · Cited by 31
“E XPERIMENTS Our framework is evaluated on two publicly available pedestrian trajectory datasets: The ETH [56] (including ETH and HO-TEL) and UCY [57] (including UNIV, ZARA1, and ZARA2).”
MethodOnline Motion Agreement Tracking
Wu, Z., Zhang, J., & Betke, M. (2013) · British Machine Vision Conference (BMVC) · doi:10.5244/c.27.63 · Cited by 31
Multimodal Interaction-Aware Trajectory Prediction in Crowded Space
Shi, X., Shao, X., Fan, Z., Jiang, R., Zhang, H., Guo, Z., Wu, G., Yuan, W., & Shibasaki, R. (2020) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v34i07.6874 · Cited by 32
“The two datasets contain 5 sets, which are UCY-zara01, UCY-zara02, UCY-univ, ETH-hotel, ETH-eth in 4 crowded scenarios with totally 1536 trajectories.”
BackgroundMethodInfluentialZhou, R. H., Zhou, H.-Y., Gao, H., Tomizuka, M., Li, J., & Xu, Z. (2022) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra46639.2022.9811585 · Cited by 33
“This is because the ETH test set contains only 2 timesteps with at least 5 people in the scene, out of the total 1161 timesteps.”
MethodInfluentialTsao, L.-W., Wang, Y., Lin, H.-S., Shuai, H.-H., Wong, L.-K., & Cheng, W.-H. (2022) · Lecture notes in computer science · doi:10.1007/978-3-031-20047-2_14 · Cited by 31
“We perform the comparisons on the benchmark datasets: ETH [31], UCY [16], and SDD [32].”
MethodAhmed, S., Al‐Bazi, A., Saha, C., Rajbhandari, S., & Huda, M. N. (2023) · Expert Systems with Applications · doi:10.1016/j.eswa.2023.120077 · Cited by 31
Static-dynamic global graph representation for pedestrian trajectory prediction
Zhou, H., Yang, X., Fan, M., Huang, H. Y., Ren, D., & Xia, H. (2023) · Knowledge-Based Systems · doi:10.1016/j.knosys.2023.110775 · Cited by 32
Measuring the Effectiveness of Readability for Mobile Robot Locomotion
Carton, D., Olszowy, W., & Wollherr, D. (2016) · International Journal of Social Robotics · doi:10.1007/s12369-016-0358-7 · Cited by 31
Chen, X., Luo, F., Zhao, F., et al. (2023) · IEEE Robotics and Automation Letters · doi:10.1109/lra.2023.3331651 · Cited by 30
“The ETH/UCY dataset [27] was collected from various locations, including squares, sidewalks, and pedestrian bridges in the city center of Zurich (ETH dataset) and different areas within the campus of the University of California, Berkeley (UCY dataset).”
Q‐EANet: Implicit social modeling for trajectory prediction via experience‐anchored queries
Chen, J., Wang, Z., Wang, J., & Cai, B. (2023) · IET Intelligent Transport Systems · doi:10.1049/itr2.12477 · Cited by 30
Chen, W., Sang, H., Wang, J., et al. (2023) · The Journal of Supercomputing · doi:10.1007/s11227-023-05850-8 · Cited by 30
“Our proposed STIGCN model is trained on two classical but still challenging data-sets, ETH [47] and UCY [48].”
Shao, R., Yang, C. H., Li, Q., Xu, L., Yang, X., Li, X., Li, M., Zhu, Q., Zhang, Y., Li, Y., et al. (2025) · IEEE Transactions on Geoscience and Remote Sensing · doi:10.1109/tgrs.2025.3526725 · Cited by 45
“Trajectory: The ETH-UCY [68], [69] dataset is a widely used benchmark for pedestrian trajectory prediction and is divided into five subsets: ETH, HOTEL, UNIV, ZARA1, and ZARA2.”
Joint Individual-Group Modeling for Tracking
Bazzani, L., Zanotto, M., Cristani, M., et al. (2014) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2014.2353641 · Cited by 30
“To further assess the performance of the proposed framework, we considered also the BIWI dataset [71].”
InfluentialTracking Persons-of-Interest via Unsupervised Representation Adaptation
Zhang, S., Huang, J., Lim, J., Gong, Y., Wang, J., Ahuja, N., & Yang, M. (2019) · International Journal of Computer Vision · doi:10.1007/s11263-019-01212-1 · Cited by 31
Collective states and their transitions in football
Welch, M., Schaerf, T. M., & Murphy, A. J. (2021) · PLoS ONE · doi:10.1371/journal.pone.0251970 · Cited by 30
Human motion prediction for intelligent construction: A review
Xia, X., Zhou, T., Du, J. K., & Li, N. (2022) · Automation in Construction · doi:10.1016/j.autcon.2022.104497 · Cited by 33
Robot Local Navigation with Learned Social Cost Functions
Pérez-Higueras, N., Ramón-Vigo, R., Caballero, F., & Merino, L. (2014) · International Conference on Informatics in Control, Automation and Robotics (ICINCO) · doi:10.5220/0005120806180625 · Cited by 30
“As a source of examples, the BIWI Walking Pedestrians dataset 2 has been used (Pellegrini et al., 2009).”
MethodAnticipatory robot path planning in human environments
Cosgun, A., Sisbot, E. A., & Christensen, H. I. (2016) · IEEE RO-MAN · doi:10.1109/roman.2016.7745174 · Cited by 30
“Pellegrini [23] trains a dynamic social behavior, that account for social interactions, using pedestrian data.”
BackgroundPredicting Pedestrian Trajectories Using Velocity-Space Reasoning
Kim, S., Guy, S. J., Liu, W., et al. (2013) · Springer tracts in advanced robotics · doi:10.1007/978-3-642-36279-8_37 · Cited by 30
“Pellegrini et al. [22] use an energy function to build up a goal-directed short-term collision avoidance motion model and Linear Trajectory Avoidance to improve their people tracking quality from video.”
BackgroundMethodInfluentialJoint Attention in Driver-Pedestrian Interaction: from Theory to PracticePreprint
Rasouli, A., & Tsotsos, J. K. (2018) · arXiv (Cornell University) · doi:10.48550/arxiv.1802.02522 · Cited by 41
“Type Data Type Cam Pose LTA [612] 2009 PP, PV,G, SC GD Traj Vid+Col BeV+FiP Early-et [598] 2012 PPs SVM Cross Img+Col F+FiP IAPA [608] 2013 PP,PV,G MDP Traj,Cross Vid+Col F Evasive [614] 2013 PPs,MH SVM Cross St+Vid+Gr F+FiP Early-Pred [621] 2014 PP,PV SVM Traj Vid+Gr Mult+FiP Veh-Pers [597] 2014 PP,PV,VD BN Traj Vid+…”
BackgroundMethodSIF-TF: A Scene-Interaction fusion Transformer for trajectory prediction
Gao, F., Huang, W.-J., Weng, L.-B., & Zhang, Y.-M. (2024) · Knowledge-Based Systems · doi:10.1016/j.knosys.2024.111744 · Cited by 29
Social Grouping for Multi-Target Tracking and Head Pose Estimation in Video
Qin, Z., & Shelton, C. R. (2015) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2015.2505292 · Cited by 29
“Social context has been explored in a number of computer vision problems.”
BackgroundCollision-Free LSTM for Human Trajectory Prediction
Xu, K., Qin, Z., Wang, G., Huang, K., Ye, S., & Zhang, H. (2018) · Lecture notes in computer science · doi:10.1007/978-3-319-73603-7_9 · Cited by 29
DeepMoTIon: Learning to Navigate Like Humans
Hamandi, M., D’Arcy, M., & Fazli, P. (2019) · IEEE International Conference on Robot and Human Interactive Communication (RO-MAN) · doi:10.1109/ro-man46459.2019.8956408 · Cited by 29
“To evaluate the performance of DeepMoTIon when imitating humans, we conducted experiments on the ETH BIWI walking pedestrians dataset [16].”
MethodContext-Aware Human Trajectories Prediction via Latent Variational Model
Berenguer, A. D., Alioscha‐Perez, M., Oveneke, M. C., & Hichem Sahli, B. (2020) · IEEE Transactions on Circuits and Systems for Video Technology · doi:10.1109/tcsvt.2020.3014869 · Cited by 30
“To assess the performance of our proposed framework we have conducted experiments on two publicly available data sets used to evaluate trajectory prediction methods in crowded environments [15]: i ) BIWI Walking Pedestrians [46] (ETH) 2 and ii ) Crowds-by-Example [16] (UCY).”
BackgroundMethodAn Online and Flexible Multi-object Tracking Framework Using Long Short-Term Memory
Wan, X., Wang, J., & Zhou, S. (2018) · IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW) · doi:10.1109/cvprw.2018.00169 · Cited by 28
“[24] [32] [25] [7] [37] [29] [1] have been combining the motion dynamics and interaction between measurements along with the target appearance.”
BackgroundForecasting Social Navigation in Crowded Complex ScenesPreprint
Robicquet, A., Alahi, A., Sadeghian, A., et al. (2016) · arXiv (Cornell University) · doi:10.48550/arxiv.1601.00998 · Cited by 29
“Similar to [27, 21], we observe trajectories for 2.4 seconds and predict for 4.8 seconds.”
BackgroundMethodInfluentialIntentQA: Context-aware Video Intent Reasoning
Li, J., Wei, P., Han, W., & Fan, L. (2023) · IEEE/CVF International Conference on Computer Vision (ICCV) · doi:10.1109/iccv51070.2023.01099 · Cited by 35
“Some studies [38, 44, 55, 52] manifest human intentions by predicting their trajectories.”
BackgroundReal Time Detection of Social Interactions in Surveillance Video
Rota, P., Conci, N., & Sebe, N. (2012) · Lecture notes in computer science · doi:10.1007/978-3-642-33885-4_12 · Cited by 28
“A novel aspect we introduce with respect to [10], consists in the insertion of an intentionality parameter in the processing chain, targeted at distinguishing between intentional and casual interactions.”
BackgroundMethodInfluentialCLUST: Simulating Realistic Crowd Behaviour by Mining Pattern from Crowd Videos
Zhao, M., Cai, W., & Turner, S. J. (2017) · Computer Graphics Forum · doi:10.1111/cgf.13259 · Cited by 28
“These real motion data can be used to calibrate the numerical parameters of crowd models as in [ABW06,JHS07,MHG ∗ 09,PPD07,POO ∗ 09,ST09].”
BackgroundMethodBighashdel, A., & Dubbelman, G. (2019) · IEEE International Conference on Intelligent Transportation Systems (ITSC) · doi:10.1109/itsc.2019.8917053 · Cited by 28
“[8] developed a dynamic social behavior based on trajectory avoidance.”
BackgroundPoibrenski, A., Klusch, M., Vozniak, I., et al. (2020) · ACM Symposium on Applied Computing · doi:10.1145/3341105.3373877 · Cited by 28
“For example, in [43] the authors propose a linear trajectory avoidance model, and in [58] the social force model is utilized.”
Multi-Object Tracking With Spatial-Temporal Topology-Based Detector
You, S., Yao, H., & Xu, C. (2021) · IEEE Transactions on Circuits and Systems for Video Technology · doi:10.1109/tcsvt.2021.3096237 · Cited by 28
Li, K., Eiffert, S., Shan, M., et al. (2021) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra48506.2021.9561480 · Cited by 28
“Deterministic Predictions: Table I (bottom) shows the results of testing the deterministic models on the ETH [44] and UCY [27] datasets, displaying both our deterministic model, and our probabilistic model that has been limited to outputting a single ‘most-likely’ prediction.”
BackgroundMethodData-driven Crowd Modeling Techniques: A Survey
Zhong, J., Li, D., Huang, Z., Lu, C., & Cai, W. (2022) · ACM Transactions on Modeling and Computer Simulation · doi:10.1145/3481299 · Cited by 33
Jin, Z., & Bhanu, B. (2015) · Computer Vision and Image Understanding · doi:10.1016/j.cviu.2014.10.001 · Cited by 29
“Similar to [11], the value of the remaining six parameters are optimized using a Genetic Algorithm [34], but the optimization is conducted on the UCSD crowd dataset [35].”
BackgroundMethodInfluentialMorales, N., Toledo, J. T., Acosta, L., & Medina, J. (2016) · IEEE Transactions on Intelligent Transportation Systems · doi:10.1109/tits.2016.2616718 · Cited by 28
“Some solutions try to model the social behavior of the pedestrians in order to improve the obtained tracks, as it happens in [28]–[30].”
BackgroundSocNav1: A Dataset to Benchmark and Learn Social Navigation Conventions
Manso, L. J., Nuñez, P., Calderita, L. V., et al. (2020) · Data · doi:10.3390/data5010007 · Cited by 28
“Two data sets are also described in [15] for tracking multiple people.”
BackgroundDERGCN: Dynamic-Evolving graph convolutional networks for human trajectory prediction
Mi, J., Zhang, X., Zeng, H., et al. (2023) · Neurocomputing · doi:10.1016/j.neucom.2023.127117 · Cited by 27
STAGP: Spatio-Temporal Adaptive Graph Pooling Network for Pedestrian Trajectory Prediction
Liu, Z., He, L., Yuan, L., Lv, K., Zhong, R., & Yao-hua, C. (2023) · IEEE Robotics and Automation Letters · doi:10.1109/lra.2023.3346806 · Cited by 27
“This section validated STAGP using ETH [43] and UCY [44] datasets.”
MethodAdapting to Length Shift: FlexiLength Network for Trajectory Prediction
Xu, Y., & Fu, Y. (2024) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr52733.2024.01442 · Cited by 33
“We conducted experiments using a Transformer-1 based method AgentFormer [73] on dataset ETH/UCY [31, 46] with results illustrated in Fig.”
MethodWei, C., Wu, G., Barth, M., Abdelraouf, A., Gupta, R. K., & Han, K. (2024) · IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW) · doi:10.1109/cvprw63382.2024.00706 · Cited by 30
“In dynamic and interactive environments, forecasting the future path of vehicles based on their current and past states becomes increasingly challenging [34].”
BackgroundData-Driven Crowd Simulation with Generative Adversarial Networks
Amirian, J., van Toll, W., Hayet, J.-B., et al. (2019) · doi:10.1145/3328756.3328769 · Cited by 27
“m (with w = 5 and smax = 2m/s), followed by the ORCA algorithm [2] as implemented by the original authors. We remove an agent when it reaches the end of its trajectory. WetestourmethodontheETH dataset[10]thatcontainsrecorded trajectories around the entrance of a university building. We have defined the region of inter…”
MethodPedestrian Trajectory Prediction via Spatial Interaction Transformer Network
Su, T., Yu, M., & Xu, Y. H. (2021) · IEEE Intelligent Vehicles Symposium Workshops (IV Workshops) · doi:10.1109/ivworkshops54471.2021.9669249 · Cited by 27
“The second is the widely used benchmark datasets ETH (ETH and HOTEL) and UCY (UNIV, ZARA1, and ZARA2) in the field of pedestrian trajectory prediction.”
MethodInfluentialZhou, H., Ren, D., Yang, X., Fan, M., & Huang, H. (2022) · Pattern Recognition · doi:10.1016/j.patcog.2022.109030 · Cited by 32
Motion planning in dynamic environments using context-aware human trajectory prediction
Finean, M. N., Petrović, L., Merkt, W., et al. (2023) · Robotics and Autonomous Systems · doi:10.1016/j.robot.2023.104450 · Cited by 27
“Physics-based methods predict human motion by propagating the current state through an explicit dynamical model [20, 61, 48, 69, 26, 22].”
MethodDROGON: A Causal Reasoning Framework for Future Trajectory Forecast.Preprint
Choi, C., Patil, A., & Malla, S. (2019) · arXiv (Cornell University) · Cited by 27
Pedestrian Intention Prediction: A Multi-task PerspectivePreprint
Bouhsain, S. A., Saadatnejad, S., & Alahi, A. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2010.10270 · Cited by 38
“Some are model-based [6, 7] which are scenario-specific and have low performance in approximating complex functions such as pedestrian trajectory.”
BackgroundHuman mobile robot interaction in the retail environment
Chen, Y., Luo, Y. G., Yang, C., Yerebakan, M. O., Hao, S., Grimaldi, N. S., Li, S., Hayes, R., & Hu, B. (2022) · Scientific Data · doi:10.1038/s41597-022-01802-8 · Cited by 27
A literature review of dense crowd simulation
Dang, H.-T., Gaudou, B., & Verstaevel, N. (2024) · Simulation Modelling Practice and Theory · doi:10.1016/j.simpat.2024.102955 · Cited by 26
Particles cross-influence for entity grouping
Rota, P., Ullah, H., Conci, N., et al. (2013) · Iris (University of Trento) · Cited by 26
Multi-target Tracking with Motion Context in Tensor Power Iteration
Shi, X., Ling, H., Hu, W., Yuan, C., & Xing, J. (2014) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr.2014.450 · Cited by 26
“Further, with the embedding of interaction based motion model, most approaches [1, 15, 14, 21] are limited to the predictive tracking framework, such as recursive filters.”
BackgroundMethodMulti-modal Multimedia Big Data Analyzing Architecture and Resource Allocation on Cloud Platform
Jayasena, K. P. N., Li, L., & Xie, Q. (2017) · Neurocomputing · doi:10.1016/j.neucom.2016.11.077 · Cited by 26
Sparse motion fields for trajectory prediction
Barata, C., Nascimento, J. C., Lemos, J. M., et al. (2020) · Pattern Recognition · doi:10.1016/j.patcog.2020.107631 · Cited by 26
Peng, H., Tsai, A.-H., Wang, L., et al. (2021) · IEEE Transactions on Cognitive Communications and Networking · doi:10.1109/tccn.2021.3115765 · Cited by 26
Zhang, C., Berger, C., & Dozza, M. (2021) · IEEE Intelligent Vehicles Symposium · doi:10.1109/iv48863.2021.9575958 · Cited by 26
“…Waymo Open Dataset for a systematic and experimental evaluation of our approach as this dataset contains more urban traffic scenarios that are nearly twice as densely populated by pedestrians than the datasets ETH [5] and UCY [6], which were mainly used for performance comparisons so far.”
BackgroundMethodVision-based Intention and Trajectory Prediction in Autonomous Vehicles: A Survey
Teeti, I., Khan, S. H., Shahbaz, A., et al. (2022) · International Joint Conference on Artificial Intelligence · doi:10.24963/ijcai.2022/785 · Cited by 26
“On the other hand, ETH [Pellegrini et al. , 2009] contemplated trajectory prediction for off-road pedestrians only.”
BackgroundZhu, D., Zhai, G., Di, Y., Manhardt, F., Berkemeyer, H., Tran, T. A., Navab, N., Tombari, F., & Busam, B. (2023) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr52729.2023.00533 · Cited by 27
“However, such marginal approximation is inherently sub-optimal [33], resulting in unsatisfactory results on realworld benchmarks [3,8,23,24,31].”
BackgroundGroup Tracking: Exploring Mutual Relations for Multiple Object Tracking
Duan, G., Ai, H., Cao, S., et al. (2012) · Lecture notes in computer science · doi:10.1007/978-3-642-33712-3_10 · Cited by 26
“[10] modeled human interactions to some degree and introduced a dynamic social behaviors to facilitate multi-people tracking.”
BackgroundChanges in Usage of an Indoor Public Space: Analysis of One Year of Person Tracking
Brščić, D., & Kanda, T. (2014) · IEEE Transactions on Human-Machine Systems · doi:10.1109/thms.2014.2374172 · Cited by 26
“Publicly available pedestrian tracking datasets that span multiple days include the MIT dataset [18] and the BIWI dataset [19].”
BackgroundTransferring human navigation behaviors into a robot local planner
Ramón-Vigo, R., Pérez-Higueras, N., Caballero, F., & Merino, L. (2014) · IEEE RO-MAN · doi:10.1109/roman.2014.6926347 · Cited by 26
“As a source of examples on which the pedestrians motion is extracted, The BIWI Walking Pedestrians dataset 2 [17] has been used (see Fig.”
MethodHybrid 3D–2D human tracking in a top view
Migniot, C., & Ababsa, F. (2014) · Journal of Real-Time Image Processing · doi:10.1007/s11554-014-0429-7 · Cited by 26
“In crowded scenarios, Pellegrini [31] modeled people social behavior to more accurately predict their movement.”
AGORASET: a dataset for crowd video analysis
Allain, P., Courty, N., & Corpetti, T. (2012) · HAL (Le Centre pour la Communication Scientifique Directe) · Cited by 26
Improving Point Cloud Semantic Segmentation by Learning 3D Object Detection
Unal, O., Gool, L. V., & Dai, D. (2021) · IEEE Winter Conference on Applications of Computer Vision · doi:10.1109/wacv48630.2021.00299 · Cited by 29
“For example cars are likely to leave substantial space between themselves and surrounding buildings; cars, cyclists and pedestrians are more likely to be rotated along the direction of the road or sidewalks respectively and away from paths of direct collision [23].”
BackgroundCausal Intervention for Human Trajectory Prediction with Cross Attention Mechanism
Ge, C., Song, S., & Huang, G. M. (2023) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v37i1.25142 · Cited by 27
“The social motion patterns, social interactions etc. are different (Amirian et al. 2020; Pellegrini et al. 2009): e.g. , people tend to form groups in the crowded UNIV domain since the scene is crowded, while pedestrians mostly keep a social distance from each other in HOTEL domain(Pellegrini, Ess,…”
BackgroundMethodEvaluating Pedestrian Trajectory Prediction Methods With Respect to Autonomous Driving
Uhlemann, N., Fent, F., & Lienkamp, M. (2024) · IEEE Transactions on Intelligent Transportation Systems · doi:10.1109/tits.2024.3386195 · Cited by 25
“…through simple rule-based methods [10], various architectures based on neural networks [11], [12], [4] as well as hybrid approaches [5], [6] have emerged in recent years, trying to improve the overall accuracy of the predicted paths on widely-used benchmarks like ETH/UCY [13], [14] and SDD [15].”
MethodKarim, M. M., Shi, Y., Zhang, S., Wang, B., Nasri, M., & Wang, Y. (2025) · Artificial Intelligence for Transportation · doi:10.1016/j.ait.2025.100004 · Cited by 35
Tracking Social Groups Within and Across Cameras
Solera, F., Calderara, S., Ristani, E., Tomasi, C., & Cucchiara, R. (2016) · IEEE Transactions on Circuits and Systems for Video Technology · doi:10.1109/tcsvt.2016.2607378 · Cited by 25
“Eventually, stu003 [32] and eth [33] sequences are good examples of single camera, mildly crowded sequences with both individual tracking and groups annotation—but are of no help in the multicamera setting.”
BackgroundVisual Exposes You: Pedestrian Trajectory Prediction Meets Visual Intention
Zhong, X., Xu, Y., Yang, Z., et al. (2023) · IEEE Transactions on Intelligent Transportation Systems · doi:10.1109/tits.2023.3266762 · Cited by 25
“craft functions [8], [9] to solve this task, but it is not capable of handling such complicated relationships and understanding various social behaviors among humans.”
InfluentialUnderstanding Dynamic Social Grouping Behaviors of Pedestrians
Feng, L., & Bhanu, B. (2014) · IEEE Journal of Selected Topics in Signal Processing · doi:10.1109/jstsp.2014.2365765 · Cited by 25
“The relationship between existing nodes and the new node is measured by the edge weights based on the spatio-temporal interactions of the tracklets.”
Background
Metadata
- Authors
- Pellegrini, S., Ess, A., Schindler, K., & van Gool, L.
- Affiliations
- ETH Zurich (CH)
- Technische Universität Darmstadt (DE)
- Venue
- 2009 IEEE 12th International Conference on Computer Vision
- Publisher
- IEEE
- Published
- 2009-09
- Pages
- 261-268
- OpenAlex
- W2532516272
- Semantic Scholar
- 7d7a98a81715e2d3747071722b2deeed8937d122
- Cited by
- 1,683
- References
- 25
- Citations by year
- 2026: 85 · 2025: 185 · 2024: 192 · 2023: 163 · 2022: 153 · 2021: 156 · 2020: 141 · 2019: 100 · 2018: 66 · 2017: 68 · 2016: 86 · 2015: 69 · 2014: 75 · 2013: 54 · 2012: 42
- FWCI
- 24.50
- Type
- conference paper
- Language
- en
- Open access
- closed
- Locations
- IEEE/CVF International Conference on Computer Vision (ICCV) (publishedVersion)
- Lirias (KU Leuven) (submittedVersion)
- Funders
- Else Kröner-Fresenius-Stiftung
- Topics
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Video Surveillance and Tracking Methods
- Physical Sciences › Engineering › Automotive Engineering › Autonomous Vehicle Technology and Safety
- Social Sciences › Safety Research › Evacuation and Crowd Dynamics
- Keywords
- multi-target tracking, social behavior modeling, crowd simulation, collision avoidance, occlusion handling, vehicle-mounted camera, bird's-eye view, data association, human trajectory prediction, social interaction
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
