Everybody needs somebody: Modeling social and grouping behavior on a linearprogramming multiple people trackerLaura Leal-Taixe, Gerard Pons-Moll and Bodo RosenhahnICCV2011

OutlineGoalMultiple people trackingModeling social behaviorExperimental resultsConclusion

GoalPeople detection is not always correct.It is important to merge the detection results into right trajectoies.

Multiple people trackingdivided in two stepsobject detectiondata associationform complete trajectoriesBuild a graph with the nodes pedestrian detectionsThe matching problem is equivalent to minimum-cost network flow problem

Multiple people tracking ,trajectory of kFind the that best explains the detection. 4

P(oi|T) is the likelihood.

Multiple people tracking trajectory Tk have following dependenciesConstant velocity assumption find oi depends on oi-1,oi-2Grouping behavior Avoidance term

Multiple people tracking

Represent by Markov chain:

Multiple people tracking

Multiple people trackingCombine (1),(2),(3)

Multiple people trackingThree kinds of edges:Link edgesDetection edgesEntrance and exit edges

Multiple people trackingLink edgesThe edges (ei, bj) connect the end nodes ei with the beginning nodes bj in following frames,with cost Ci,j and flag fi,jFlag =1 if oi and oj belong to Tk,and fFmax111

Multiple people trackingDetection edgesThe edges (bi, ei) connect the beginning node bi and end node ei, with cost Ci and flag fi

Modeling social behaviorIf a pedestrian doesnt meet any obstacles, he will naturally follow a straight line.But the pedestrian will have some social behavior.Add Social Force Model (SFM)and Group behavior(GR) into the problem.

Modeling social behaviorSocial forces have three main terms:The desire to maintain certain speedThe desire to keep away from othersThe desire to reach a destinationWe focus on first two!

Modeling social behavior

Constant velocity assumpionWhen a person walk at a speed V at time tWe assume he will have speed V at time t+t

Modeling social behaviorAvoidance term

Modeling social behaviorFrom the training sequence in [22] , we learn the probabilty of Pg and Pi

[22] S. Pellegrini, A. Ess, K. Schindler, and L. van Gool. Youll never walk alone: modeling social behavior for multi-target tracking. ICCV, 2009. 1, 2, 5, 7

Experimental resultsBlue=>DISTGreed=>with SDMRed=>SFM+GR

Experimental results

Experimental resultsTo show the importance of social behavior and the robustness of our algorithm at low frame rates, we track at 2.5fps (taking one every tenth frame).

Experimental resultsDA (detection accuracy)TA (tracking accuracy)DP (detection precision)TP (tracking precision)

Experimental results

[28]use network flow[22]use social behavior[27] use social and grouping

Experimental results

ConclusionIt is important to have social and group relation on tracking.This paper outperform on low fps than others and have high accuracies on miss detections,false alarms and noise.