A Unified Framework for Multi-target Tracking and Collective Activity Recognition

By Wongun Choi, Silvio Savarese. European Conference on Computer Vision, 2012

378 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition

View Paper: https://www.semanticscholar.org/paper/f5900377764ab70f1fad70a5fb42148e864c6810

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

3D Traffic Scene Understanding From Movable Platforms (2014, neutral): “Conclusions are drawn in Section 6.”
Learning Social Etiquette: Human Trajectory Understanding In Crowded Scenes (2016, methodology): “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”
Learning by Tracking: Siamese CNN for Robust Target Association (2016, background): “A recent trend is to design ever more complex models, which include further vision routines in the hope that they benefit the tracker, including reconstruction for multicamera sequences [36, 54], activity recognition [11] and segmentation [40].”
Recurrent Modeling of Interaction Context for Collective Activity Recognition (2017, methodology): “We follow the train/test splits as suggested in [22], use the tracklet data provided in [5] and only use five group-level activities for training.”
Video-Based Human Behavior Understanding: A Survey (2013, background): “Examples include action [77], [105], [174], and interaction analysis [161], [175].”
On pairwise costs for network flow multi-object tracking (2014, methodology): “1, 2, 6 [9] W. Choi and S. Savarese.”
Social LSTM: Human Trajectory Prediction in Crowded Spaces (2016, methodology): “This method was later extended to robotics [41] and activitiy understanding [43, 73, 50, 38, 37, 9, 10].”
Spatio-Temporal Dynamic Inference Network for Group Activity Recognition (2021, neutral): “Following works [10, 25, 24, 8, 2] were basically to extract hand-crafted features ( e.g. , HOG [12]) and apply graphical models to infer group activity representations.”
Structure Inference Machines: Recurrent Neural Networks for Analyzing Relations in Group Activity Recognition (2015, methodology): “Many previous work have used various forms of graphical models to address this problem: hierarchical graphical models [2, 24, 25, 27, 9], ANDOR graphs [3, 16] and dynamic Bayesian networks [35] are among the popular models.”
Deep Network Flow for Multi-object Tracking (2017, methodology): “In particular, structured SVMs [48] have recently been used in the tracking context to learn costs for bipartite graph matching in an on-line tracker [23], a divide-and-conquer tracking strategy [46] and a joint graphical model for activity recognition and tracking [12].”
D Traffic Scene Understanding from Movable Platforms (2013, background): “[15] introduce a hierarchy of activities, modeling the behavior of groups.”
Tracking Without Bells and Whistles (2019, background): “A recent trend is to design ever more complex models which include other vision input such as reconstruction for multi-camera sequences [41, 57], activity recognition [14], segmentation [45], keypoint trajectories [12] or joint detection [56].”
Hierarchical Deep Temporal Models for Group Activity Recognition (2016, background): “Choi and Savarese [21] unified tracking multiple people, recognizing individual actions, interactions and collective activities in a joint framework.”
Learning to Detect Partially Overlapping Instances (2013, background): “an image from a surveillance camera may contain multiple individual pedestrians but also few groups of people which are hard to segment from each other [7].”
Multiple object tracking: A literature review (2014, methodology): “2009; Andriyenko and Schindler 2010), min-cost max-flow network flow (Zhang et al. 2008; Choi and Savarese 2012; Wu et al. 2012; Butt and Collins 2013; Pirsiavash et al. 2011) to max weight independent set (Shafique et al.”
Learning a Neural Solver for Multiple Object Tracking (2019, background): “A recent trend is to design ever more complex models which include other vision input such as reconstruction for multi-camera sequences [43, 67], activity recognition [15], segmentation [49], keypoint trajectories [14] or joint detection [65].”
VITAL: VIsual Tracking via Adversarial Learning (2018, background): “Visual tracking has a wide range of applications including action recognition [7], target analysis [52, 51, 49] and augmented reality [8, 44].”