Learning a Pedestrian Social Behavior Dictionary
Not on the map. Why?
Abstract (arXiv)
Understanding pedestrian behavior patterns is a key component to building autonomous agents that can navigate among humans. We seek a learned dictionary of pedestrian behavior to obtain a semantic description of pedestrian trajectories. Supervised methods for dictionary learning are impractical since pedestrian behaviors may be unknown a priori and the process of manually generating behavior labels is prohibitively time consuming. We instead utilize a novel, unsupervised framework to create a taxonomy of pedestrian behavior observed in a specific space. First, we learn a trajectory latent space that enables unsupervised clustering to create an interpretable pedestrian behavior dictionary. We show the utility of this dictionary for building pedestrian behavior maps to visualize space usage patterns and for computing the distributions of behaviors. We demonstrate a simple but effective trajectory prediction by conditioning on these behavior labels. While many trajectory analysis methods rely on RNNs or transformers, we develop a lightweight, low-parameter approach and show results comparable to SOTA on the ETH and UCY datasets.
Key facts. Cited by 1 paper in the map (data through 7 Oct 2026). Builds on 29 of its 52 references that are in the map. First posted on arXiv 2 Dec 2022.
Builds on
29 of 52 references.
References · 29
- 2020Unsupervised Learning of Visual Features by Contrasting Cluster Assignmentscited by 2,845
- 2018Deep Clustering for Unsupervised Learning of Visual Featurescited by 1,020
- 2018Social GAN: Socially Acceptable Trajectories with Generative Adversarial Networkscited by 687
- 2021AgentFormer: Agent-Aware Transformers for Socio-Temporal Multi-Agent Forecastingcited by 327
- 2020Social-STGCNN: A Social Spatio-Temporal Graph Convolutional Neural Network for Human Trajectory Predictioncited by 324
- 2018SoPhie: An Attentive GAN for Predicting Paths Compliant to Social and Physical Constraintscited by 296
- 2020It Is Not the Journey but the Destination: Endpoint Conditioned Trajectory Predictioncited by 235
- 2019Social-BiGAT: Multimodal Trajectory Forecasting using Bicycle-GAN and Graph Attention Networkscited by 228
- 2020Spatio-Temporal Graph Transformer Networks for Pedestrian Trajectory Predictioncited by 218
- 2017Social Attention: Modeling Attention in Human Crowdscited by 185
- 2020Transformer Networks for Trajectory Forecastingcited by 180
- 2020Human Trajectory Forecasting in Crowds: A Deep Learning Perspectivecited by 128
17 more
- 2022GroupNet: Multiscale Hypergraph Neural Networks for Trajectory Prediction with Relational Reasoningcited by 128
- 2020Learning Representations by Predicting Bags of Visual Wordscited by 73
- 2020Recursive Social Behavior Graph for Trajectory Predictioncited by 66
- 2022SocialVAE: Human Trajectory Prediction using Timewise Latentscited by 65
- 2022MUSE-VAE: Multi-Scale VAE for Environment-Aware Long Term Trajectory Predictioncited by 59
- 2020Social NCE: Contrastive Learning of Socially-aware Motion Representationscited by 58
- 2021Three Steps to Multimodal Trajectory Prediction: Modality Clustering, Classification and Synthesiscited by 46
- 2016Deep TEN: Texture Encoding Networkcited by 44
- 2021Interpretable Social Anchors for Human Trajectory Forecasting in Crowdscited by 44
- 2022Learning Pedestrian Group Representations for Multi-modal Trajectory Predictioncited by 44
- 2018Deep Texture Manifold for Ground Terrain Recognitioncited by 32
- 2021Personalized Trajectory Prediction via Distribution Discriminationcited by 30
- 2022Goal-driven Self-Attentive Recurrent Networks for Trajectory Predictioncited by 28
- 2019Social and Scene-Aware Trajectory Prediction in Crowded Spacescited by 19
- 2020AC-VRNN: Attentive Conditional-VRNN for Multi-Future Trajectory Predictioncited by 16
- 2020Unsupervised Learning of Video Representations via Dense Trajectory Clusteringcited by 11
- 2020Pedestrian Motion Model Using Non-Parametric Trajectory Clustering and Discrete Transition Pointscited by 4
Built on by
Citing papers per year: 2023: 1.
1 paper.