Coupled Detection and Trajectory Estimation for Multi-Object Tracking

By Bastian Leibe, Konrad Schindler, Luc Van Gool. IEEE International Conference on Computer Vision, 2007

318 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition

View Paper: https://www.semanticscholar.org/paper/6d0db7dc45147d5d5680a0918fc1a1e23b0124d2

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

Citations in the corpus

GMCP-Tracker: Global Multi-object Tracking Using Generalized Minimum Clique Graphs (2012, methodology): “The approach in [6] first generates low level tracklets and then employs heuristics in order to merge tracklets into trajectories.”
Learning People Detectors for Tracking in Crowded Scenes (2013, methodology): “In [13], the task is formulated as a quadratic Boolean program to combine trajectory estimation and detection.”
Multi-target tracking by learning local-to-global trajectory models (2015, result): “In contrast to the local tracking methods, many of the latter approaches [20–22,7,23,24] have achieved great progress by using global inference over all trajectories simultaneously in a longer period.”
People-tracking-by-detection and people-detection-by-tracking (2008, background): “In that, this paper is also related to multiplepeople blob-tracking methods, such as [11], but we do not need to assume a static camera and allow for low viewpoints (also in contrast to [15]) from which people can fully occlude each other.”
Continuous Energy Minimization for Multitarget Tracking (2014, methodology): “However, to keep the computation tractable, the solution space is restricted to a finite number of states, which is usually done by only allowing trajectories to pass exactly through either non-maxima suppressed object detections [24], [27], [43], or through locations on a regular discrete grid [3],”
Robust Multiperson Tracking from a Mobile Platform (2009, background): “…on the authors, the rather loose notion of " context " can refer to different types of complementary information, including motion [11], [54], stereo depth [15], [19], scene geometry [24], [29], temporal continuity [1], [31], [32], [57], or semantics of other image regions [36], [40], [50], [51].”
Global data association for multi-object tracking using network flows (2008, methodology): “Also in [10], detection and estimation of trajectory hypotheses are coupled by Quadratic Boolean Programming.”
Coupled Object Detection and Tracking from Static Cameras and Moving Vehicles (2008, background): “However, as shown in [30], the two tasks are closely coupled and their results can mutually reinforce each other.”
Globally-optimal greedy algorithms for tracking a variable number of objects (2011, methodology): “A popular approach to multi-object tracking is to run a low-level tracker to obtain “tracklets”, and then stitch together tracklets using various graph-based formalisms or greedy heuristics [15, 22, 16, 19, 2].”
Privacy preserving crowd monitoring: Counting people without people models or tracking (2008, background): “This is a canonical example of a problem that vision technology addresses with privacy invasive methods: detect the people in the scene [3, 4, 5, 6, 7], track them over time [8, 9, 10], and count the number of tracks.”
Counting People With Low-Level Features and Bayesian Regression (2012, background): “Yet, it is frequently addressed with “individual-centric” methods: detect the people in the scene [1]–[6], track them over time [3], [7]–[9], and count the number of tracks.”
Learning to associate: HybridBoosted multi-target tracker for crowded scene (2009, methodology): “[4][5][6][7][8] aim at simultaneously optimizing all tracks in polynomial time by tailoring various optimization algorithms such as Linear Programming [4], Quadratic Boolean Programming [5], min-cost flow [6] and Hungarian algorithm [7] or its hierarchical version [8].”
Object localization in ImageNet by looking out of the window (2015, methodology): “These annotations could be used as training data for problems such as object class detection [8], tracking [21] and pose estimation [4].”
Robust tracking-by-detection using a detector confidence particle filter (2009, methodology): “Several recent multi-object tracking-by-detection algorithms address the resulting data association problem by optimizing detection assignments over a larger temporal window [1, 3, 12, 16, 21].”
Stable multi-target tracking in real-time surveillance video (2011, background): “The second group covers data association based methods which also use future information to estimate the current state, allowing ambiguities to be more easily resolved at the cost of increased latency [13, 11, 12, 3, 20].”
Multiple Object Tracking Using K-Shortest Paths Optimization (2011, unknown): “Leibe et al. [30] simultaneously optimize detections and tracks, coupled into a Quadratic Boolean Problem and solved by an EM algorithm.”
Multi-target tracking by continuous energy minimization (2011, methodology): “The resulting optimization problems are either quadratic integer programs [7, 11], in which case they are solved to local optimality by custom heuristics based on recursive search or graph cuts; or integer linear programs (ILP) [2, 4, 10], which are solved to near-global optimality through LPrelaxat”