• Corpus ID: 16967679

Temporal and Object Relations in Unsupervised Plan and Activity Recognition

  title={Temporal and Object Relations in Unsupervised Plan and Activity Recognition},
  author={Richard Gabriel Freedman and Hee-Tae Jung and Shlomo Zilberstein},
  booktitle={AAAI Fall Symposia},
We consider ways to improve the performance of unsupervised plan and activity recognition techniques by considering temporal and object relations in addition to postural data. Temporal relationships can help recognize activities with cyclic structure and are often implicit because plans have degrees of ordering actions. Relations with objects can help disambiguate observed activities that otherwise share a user’s posture and position. We develop and investigate graphical models that extend the… 

Figures and Tables from this paper

Does the Human's Representation Matter for Unsupervised Activity Recognition?
Several ways that motion capture information can be represented for use in unsupervised activity recognition methods are presented, and it is illustrated how the representation choice has the potential to produce variations in the learned clusters.
Learning features combination for human action recognition from skeleton sequences
Integrating Planning and Recognition to Close the Interaction Loop
A framework that integrates these processes by taking advantage of features shared between them is proposed, which can be risky in real-time situations where there may be enough time to only run a few steps.
Hierarchical Dual Attention-Based Recurrent Neural Networks for Individual and Group Activity Recognition in Games
This work proposes a novel framework by developing a hierarchical dual attention RNN-based method that leverages feature and temporal attention mechanisms in a hierarchical setting for effective discovery of activities using interactions among individuals.
Using Metadata to Automate Interpretations of Unsupervised Learning-Derived Clusters
This work presents an approach to assist human verification of the unsupervised learning algorithms’ classification choices through the use of metadata describing the inputs to be clustered and shows how a similar measurement of relevance to human-interpretable features can be derived to describe the un supervised learning algorithm’s choices of clusters.
Representation, Use, and Acquisition of Affordances in Cognitive Systems
This work considers issues that arise in representing mental affordances, using them to understand and generate plans, and learning them from experience, and presents theoretical claims that form an incipient theory of affordance in cognitive systems.
Human modeling for human–robot collaboration
The many techniques available for modeling human cognition and behavior are discussed, compared and contrast, and their benefits and drawbacks in the context of human–robot collaboration are evaluated.
Improving Computer Network Operations Through Automated Interpretation of State
This research presents a novel approach called “Smart grids” that automates the very labor-intensive and therefore time-heavy and expensive and therefore expensive process of designing and implementing smart grids.


Plan and Activity Recognition from a Topic Modeling Perspective
The application of Latent Dirichlet Allocation topic models to human skeletal data of plan execution traces obtained from a RGB-D sensor is explored and initial empirical results suggest that such NLP methods can be useful in complex PR and AR tasks.
Learning Spatio-Temporal Structure from RGB-D Videos for Human Activity Detection and Anticipation
This paper proposes a graph structure that improves the state-of-the-art significantly for detecting past activities as well as for anticipating future activities, on a dataset of 120 activity videos collected from four subjects.
Plan, Activity, and Intent Recognition: Theory and Practice
Plan recognition, activity recognition, and intent recognition together combine and unify techniques from user modeling, machine vision, intelligent user interfaces, human/computer interaction,
Analyzing Team Actions with Cascading HMM
This paper uses Cascading Hidden Markov Models (CHMM) to analyze Bounding Overwatch, an important team action in military tactics, and investigates whether the better scalability and the more structured information provided by the CHMM comes with an unacceptable cost in accuracy.
Activity Discovery and Activity Recognition: A New Partnership
This paper describes a method by which activity discovery can be used to identify behavioral patterns in observational data and demonstrates that activity discovery not only sheds light on behavioral patterns, but it can also boost the performance of recognition algorithms.
4-dimensional local spatio-temporal features for human activity recognition
  • Hao Zhang, L. Parker
  • Computer Science
    2011 IEEE/RSJ International Conference on Intelligent Robots and Systems
  • 2011
A new 4-dimensional (4D) local spatio-temporal feature that combines both intensity and depth information that is suitable for the task of human activity recognition is proposed.
Feature Set Selection and Optimal Classifier for Human Activity Recognition
A thorough analysis of features and classifiers aimed at human activity recognition is presented, based on a set of 10 activities, and the interdependency between feature selection method and chosen classifier is investigated.
Unstructured human activity detection from RGBD images
This paper uses a RGBD sensor as the input sensor, and compute a set of features based on human pose and motion, as well as based on image and point-cloud information, based on a hierarchical maximum entropy Markov model (MEMM).
Understanding human intentions via Hidden Markov Models in autonomous mobile robots
This paper proposes an approach that allows a robot to detect intentions of others based on experience acquired through its own sensory-motor capabilities, then using this experience while taking the perspective of the agent whose intent should be recognized.