What are they doing? : Collective activity classification using spatio-temporal relationship among people

By Wongun Choi, Khuram Shahid, Silvio Savarese. 2009 IEEE 12th International Conference on Computer Vision Workshops, ICCV Workshops, 2009

373 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition

View Paper: https://www.semanticscholar.org/paper/8d5f8d63a51dbb1554bcab132dd87ac911d9f6ec

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

Hierarchical Deep Temporal Models for Group Activity Recognition (2016, methodology): “First, we demonstrate performance on the Collective Activity Dataset [1], a surveillance-type video dataset.”
Learning Social Etiquette: Human Trajectory Understanding In Crowded Scenes (2016, background): “Researchers have demonstrated that it is indeed possible to model the interaction between humans and their surroundings to improve or solve numerous computer vision tasks: for instance, to make pedestrian tracking more robust and accurate [1,2,3,4,5], to enable the understanding of activities perfor”
A Survey on Vision Transformer (2020, methodology): “In addition to human action recognition, transformer has been utilized for group activity recognition [42].”
Evaluation of video activity localizations integrating quality and quantity measurements (2014, background): “Several datasets feature crowd behavior, for instance PETS 2009 [29], ETISEO [20], or group activities, for instance BEHAVE [23] and Collective Activity [30].”
Structure Inference Machines: Recurrent Neural Networks for Analyzing Relations in Group Activity Recognition (2015, background): “As per standard arguments around context [32, 12, 10, 25], using the actions of other people in the scene can help to disambiguate the action of this individual.”
Collective Classification in Network Data (2008, background): “This approach has been used in a number of publications [7,16] to achieve current state-of-the-art performance on benchmark datasets.”
Spatio-Temporal Dynamic Inference Network for Group Activity Recognition (2021, neutral): “Datasets So far, there are two widely used datasets in group activity recognition, namely Volleyball dataset (VD) [22] and Collective Activity dataset (CAD) [9].”
Who are you with and where are you going? (2011, background): “Social interaction in the pedestrian group began to be studied only recently in computer vision [2, 3, 9].”
Recurrent Modeling of Interaction Context for Collective Activity Recognition (2017, methodology): “We perform extensive experiments on the Collective Activity Dataset [6] and Choi’s new Dataset [5] to validate the ability of learning context information, also compare our results with the state-of-the-art methods.”
Video-Based Human Behavior Understanding: A Survey (2013, background): “dataset [103] with actions, such as standing, queuing, walking, and talking, they obtain an average accuracy of 82%.”
LEMMA: A Multi-view Dataset for Learning Multi-agent Multi-task Activities (2020, background): “Another related stream of work is the learning of group-level activities in a multi-agent setting [24], such as detecting key actors [40], predicting future trajectories [38,32], and recognizing collective activities [10,37,46].”
Improving Data Association by Joint Modeling of Pedestrian Trajectories and Groupings (2010, methodology): “Applications include simulation [2–4], computer graphics [5, 6], and, in the last few years, also Computer Vision [7–14].”
Discriminative Latent Models for Recognizing Contextual Group Activities (2012, background): “The work of Jain et al. [7] is close to our work in spirit, which also uses a learned structure instead of a fully connected model.”
A Unified Framework for Multi-target Tracking and Collective Activity Recognition (2012, methodology): “Collective activity estimation is helped by observations OC , which use features such as spatio-temporal local descriptors [1, 2] to encode the flow of people around individuals.”
CERN: Confidence-Energy Recurrent Network for Group Activity Recognition (2017, methodology): “The Collective Activity dataset consists of 44 videos, annotated with 5 activity categories (crossing, walking, waiting, talking, and queueing), 6 individual action labels (NA, crossing, walking, waiting, talking, and queueing), and 8 pairwise interaction labels (NA, approaching, leaving, passing-by”
Beyond Actions: Discriminative Models for Contextual Group Activities (2010, methodology): “The work in [3] is the closest to ours.”