You'll never walk alone: Modeling social behavior for multi-target tracking

By Stefano Pellegrini, Andreas Ess, Konrad Schindler, Luc Van Gool. IEEE International Conference on Computer Vision, 2009

1,809 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition, Ocean Engineering

View Paper: https://www.semanticscholar.org/paper/7d7a98a81715e2d3747071722b2deeed8937d122

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Computer Vision and Pattern Recognition

In Ocean Engineering

Citations in the corpus

Tracking the Untrackable: Learning to Track Multiple Cues with Long-Term Dependencies (2017, methodology): “Many models are based on raw pixel template representation for simplicity [71, 4, 68, 51, 50], while color histogram is the most popular representation for appearance modeling in MOT approaches [9, 35].”
DESIRE: Distant Future Prediction in Dynamic Scenes with Interacting Agents (2017, background): “Predicting the dynamics of multiple objects is also studied in [24, 25, 3, 31], to name a few.”
SR-LSTM: State Refinement for LSTM Towards Pedestrian Trajectory Prediction (2019, methodology): “There are two types of metrics for evaluating the performance of trajectory prediction, including the Mean Average Displacement (MAD) error and Final Average Displacement (FAD) error [31] in meters.”
PRECOG: PREdiction Conditioned on Goals in Visual Multi-Agent Settings (2019, neutral): “In contrast to the commonly-used average displacement error (ADE) and final displacement error (FDE) metrics that computes the mean Euclidean error from a batch of samples to a single ground-truth sample [1, 6, 10, 14, 28], minMSD has the desirable property of not penalizing plausible samples that c”
Multi-target tracking by learning local-to-global trajectory models (2015, methodology): “The tracking problem is solved by the data association [1–8] which strives to establish a unique identity for each target, and to simultaneously estimate the motion patterns of all targets and the assignment of detections to targets, by linking similar detections across frames.”
Human motion trajectory prediction: a survey (2019, methodology): “Motion state (position and possibly velocity), e.g. (Ferrer and Sanfeliu 2014; Elfring et al. 2014; Pellegrini et al. 2009; Kitani et al. 2012; Karasev et al. 2016; Ziebart et al. 2009; Kooij et al. 2018; Trautman and Krause 2010; Kuderer et al. 2012; Bennewitz et al. 2005; Kucner et al. 2017; Bera…”
Learning Social Etiquette: Human Trajectory Understanding In Crowded Scenes (2016, methodology): “Similar to [18] we model such obstacles as agents with zero velocity.”
TrafficPredict: Trajectory Prediction for Heterogeneous Traffic-Agents (2018, methodology): “There are some pedestrian trajectory datasets like ETH (Pellegrini et al. 2009), UCY (Lerner et al. 2007), etc., but such datasets only focus on human crowds without any vehicles.”
Robot navigation in dense human crowds: Statistical models and experimental studies of human–robot cooperation (2015, methodology): “In addition, this joint collision avoidance criteria has been used to improve the data association and target tracking of individuals in human crowds (Pellegrini et al., 2009, 2010; Luber et al., 2010).”
Improving Data Association by Joint Modeling of Pedestrian Trajectories and Groupings (2010, background): “Only some researchers use social structures to improve tracking, most notably in crowded scenarios [7], or by modeling of collision avoidance-behavior [8, 11].”
Context-Based Pedestrian Path Prediction (2014, background): “The related Linear Trajectory Avoidance model [28] for short-term path prediction uses the expected point of closest approach to foreshadow and avoid possible collisions.”
Are They Going to Cross? A Benchmark Dataset and Baseline for Pedestrian Crosswalk Behavior (2017, background): “The common approach to remedy the problem of predicting road users behavior is to employ dynamic factors, such as trajectory [3] or velocity [30], or the expected final goal of pedestrians [18, 31].”
Who are you with and where are you going? (2011, methodology): “We use the model described in [8] to capture this effect: Note that this term requires three parameters σ d , σ w , and β .”
MultiPath: Multiple Probabilistic Anchor Trajectory Hypotheses for Behavior Prediction (2019, methodology): “Our model contrasts with popular past approaches which either provide only a single MAP trajectory [1, 2, 9, 10, 11] or an unweighted set of samples via a generative model [3, 4, 6, 12, 13, 14, 15].”
UA-DETRAC: A new benchmark and protocol for multi-object detection and tracking (2015, methodology): “Early multi-object tracking methods formulate the task as the state estimation problem using Kalman (Leven and Lanter-man, 2009; Pellegrini et al., 2009) and particle ﬁlters (Isard and Blake, 1998; Khan et al., 2005; Mikami et al., 2009; Yang et al., 2014).”
Near-Online Multi-target Tracking with Aggregated Local Flow Descriptor (2015, neutral): “(e.g. A1 = {d1, d10, d23}) during the time of presence; i.e. (V T1 ,DT1 )→ A.”
Tracking Without Bells and Whistles (2019, background): “The most common assumption is the one of constant velocity (CVA) [11, 2], but pedestrian motion gets more complex in crowded scenarios for which researchers have turned to the more expressive Social Force Model [57, 48, 61, 39].”
PIE: A Large-Scale Dataset and Models for Pedestrian Intention Estimation and Trajectory Prediction (2019, background): “Trajectory-based algorithms rely on observation of pedestrians’ past motion history and predict the location of the pedestrians in the future using contextual information such as 3D depth [14, 34, 39], social interactions [25, 42, 1, 31, 41], the ego-vehicle dynamics and the scene structure [16, 17,”