Unimore logo AImageLab
Back to the research area

Group Detection and Crowd Analysis

Behavior analysis will play a central role in future video surveillance systems as research on this topic has been revealing promising in helping to discover public safety risks or predict crimes. Nevertheless, trying to understand complex interactions in the scene just by looking at each individual separately is unrealistic, due to the inherent social nature of human behavior. This is because those interactions do not occur at an individual level nor at a crowd level, but they typically involve small subsets of people, namely groups. We thus believe future challenges will reside in enhancing action analysis by considering social interactions among small gathering of people sharing a common goal, to this end group detection becomes a mandatory step for modern crowd surveillance systems.

example group detection

Pedestrian trajectories encode many sociological and physical information about the way people interact. If two pedestrians have diverging trajectories it's very unlikely that they were on the scene together and, at the same time, in a group of friends everyone will likely have very similar and compact trajectories over a generic period of observation. Starting from this consideration we reformulate the problem of finding groups in the scene as the one of clustering trajectories, or partitioning the set of those.

Besides, the spatial organization of pedestrians inside groups tends to obey to patterns that facilitate social interactions and verbal communications, while trying to avoid collisions with in-group members and out-group pedestrians. This patterns are highly dependent on the crowd density, the environment conformation and the group speed and even if it's not completely clear yet how this elements are all correlated together, we want to consider some pattern more probable than others while searching the crowd for groups.

In this research activity we developed a supervised hierarchical bottom-up correlation clustering for solving the group detection task when trajectories of pedestrians are available. Since a learning approach is followed to perform clustering, a formal definition of groups isn’t required, as social studies underlines the lack of a universal theory about their formation. A novel set of features inspired by sociology and econometric is presented and through a modified version of the Structural SVM, which let us evaluate the quality of the prediction according to their shape, we learn how to linearly combine those features in order to find a distance measure able to explain the concept of groups in any particular scenario.

Group Detection Page and Code

URL: http://imagelab.unimore.it/group-detection


1 Solera, Francesco; Calderara, Simone; Cucchiara, Rita "Socially Constrained Structural Learning for Groups Detection in Crowd" IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE, vol. 38, pp. 995 -1008 , 2016 | DOI: 10.1109/TPAMI.2015.2470658 Journal
2 Solera, Francesco; Calderara, Simone; Cucchiara, Rita "Learning to identify leaders in crowd" 2015 IEEE Conference on Computer Vision and Pattern Recognition Workshops (CVPRW), vol. 2015-, Boston USA, pp. 43 -48 , 7-12 June 2015, 2015 | DOI: 10.1109/CVPRW.2015.7301282 Conference
3 Alletto, Stefano; Serra, Giuseppe; Calderara, Simone; Solera, Francesco; Cucchiara, Rita "From Ego to Nos-Vision: Detecting Social Relationships in First-Person Views" 2014 IEEE Conference on Computer Vision and Pattern Recognition Workshops, Columbus, Ohio, 23-28 June 2014, 2014 | DOI: 10.1109/CVPRW.2014.91 Conference
4 Solera, Francesco; Calderara, Simone; Cucchiara, Rita "Structured learning for detection of social groups in crowd" 2013 10th IEEE International Conference on Advanced Video and Signal-Based Surveillance : AVSS 2013 : August 27-30, 2013, Krako´w, Poland, vol. 0, Krakov (PL), pp. 7 -12 , August 27-30 2013, 2013 | DOI: 10.1109/AVSS.2013.6636608 Conference
5 Solera, Francesco; Calderara, Simone "Social Groups Detection in Crowd through Shape-Augmented Structured LearningImage Analysis and Processing – ICIAP 2013" Lecture Notes in Computer ScienceImage Analysis and Processing – ICIAP 2013, vol. 8156, Napoli, pp. 542 -551 , 9-13 Settembre 2013, 2013 Conference

Video Demo

Research Activity Info