Showing papers on "Activity recognition published in 2001"

Recognizing human motion with multiple acceleration sensors

Abstract: In this paper experiments with acceleration sensors are described for human activity recognition of a wearable device user. The use of principal component analysis and independent component analysis with a wavelet transform is tested for feature generation. Recognition of human activity is examined with a multilayer perceptron classifier. Best classification results for recognition of different human motion were 83-90%, and they were achieved by utilizing independent component analysis and principal component analysis. The difference between these methods turned out to be negligible.

Action Based Collaboration Analysis for Group Learning

Abstract: In computer-supported collaborative learning, technology is used for enabling constructive approaches to learning as well as group interaction, particularly in distant and asynchronous situations. Recently, shared workspace systems have been developed that allow for jointly constructing conceptual problem representations by means of graphical structures and direct manipulation as well as the incorporation of existing material and the persistent storage of results. Though these systems feature free and open-ended collaboration, they lack any kind of awareness of relevant group interaction aspects. However, an explicit description of collaboration processes is necessary for monitoring and visualizing group interaction, enabling didactic interventions and intelligent support, empirically investigating human collaboration, and evaluating system design. The automatic analysis of user actions in shared workspaces (action-based collaboration analysis) is regarded as a promising new direction. In contrast to discourse-oriented approaches, it especially takes into account the operational grounding of group interactions in the context of collaboratively constructed problem representations. Based on software engineering considerations, a generic plug-in agent architecture has been developed to generally provide operational semantics and intelligent support for di erent types of user interfaces, particularly shared workspaces. The user interfaces and intelligent components communicate by broadcasting messages that represent the creation, deletion, and modi cation of objects. Realized as an intelligent component, activity recognition automatically and incrementally infers abstract notions of group activity and interpretations of problem-related con icts and coordinations from a stream of action messages. The approach, which is related to plan recognition in the situation calculus, has been formalized based on concepts of basic actions and complex, higher-level activities, situations that reconstruct relevant aspects of the user environment, operators that specify a hierarchy of activities, and pending actions that represent potential subsequent action sequences. Activity recognition does not rely on domain and task knowledge, though it can use possibly available information to improve its results. The results of the interaction analysis are visualized in di erent forms including direct feedback to the shared workspaces. The feasibility of the approach has been veri ed with a number of test subjects in realistic face-to-face sessions, with activity recognition being used for an analysis-based indexing of video recordings.

Towards a unified framework for tracking and analysis of human motion

Abstract: We propose a framework for detecting, tracking and analyzing non-rigid motion based on learned motion patterns. The framework features an appearance based approach to represent the spatial information and hidden Markov models (HMM) to encode the temporal dynamics of the time varying visual patterns. The low level spatial feature extraction is fused with the temporal analysis, providing a unified spatio-temporal approach to common detection, tracking and classification problems. This is a promising approach for many classes of human motion patterns. Visual tracking is achieved by extracting the most probable sequence of target locations from a video stream using a combination of random sampling and the forward procedure from HMM theory. The method allows us to perform a set of important tasks such as activity recognition, gait-analysis and keyframe extraction. The efficacy of the method is shown on both natural and synthetic test sequences.

Design of a digital library for human movement

Abstract: This paper is focused on a central aspect in the design of our planned digital library for human movement, i.e. on the aspect of representation and recognition of human activity from video data. The method of representation is important since it has a major impact on the design of all the other building blocks of our system such as the user interface/query block or the activity recognition/storage block. In this paper we evaluate a representation method for human movement that is based on sequences of angular poses and angular velocities of the human skeletal joints, for storage and retrieval of human actions in video databases. The choice of a representation method plays an important role in the database structure, search methods, storage efficiency etc.. For this representation, we develop a novel approach for complex human activity recognition by employing multidimensional indexing combined with temporal or sequential correlation. This scheme is then evaluated with respect to its efficiency in storage and retrieval.For the indexing we use postures of humans in videos that are decomposed into a set of multidimensional tuples which represent the poses/velocities of human body parts such as arms, legs and torso. Three novel methods for human activity recognition are theoretically and experimentally compared. The methods require only a few sparsely sampled human postures. We also achieve speed invariant recognition of activities by eliminating the time factor and replacing it with sequence information. The indexing approach also provides robust recognition and an efficient storage/retrieval of all the activities in a small set of hash tables.

View-based human activity recognition by indexing and sequencing

Abstract: A novel method for view-based recognition of human activity is presented. The basic idea of our method is that activities can be positively identified from a sparsely sampled sequence of few body poses acquired from videos. In our approach, an activity is represented by a set of pose and velocity vectors for the major body parts (hands, legs and torso) and stored in a set of multidimensional hash tables. We show that robust recognition of a sequence of body pose vectors can be achieved by a method of indexing and sequencing and it requires only few vectors (i.e. sampled body poses in video frames). We find that the probability of false alarm drops exponentially with the increased number of sampled body poses. We also achieve speed invariant recognition by eliminating the time factor and replacing it with sequence information. Experiments performed with videos having 8 different activities show robust recognition even for different viewing directions.

A smart camera for real-time human activity recognition

Abstract: This paper describes a smart camera system under development at Princeton University. This smart camera is designed for use in a smart room in which the camera detects the presence of a person in its visual field and determines when various gestures are made by the person. As a first step toward a VLSI implementation, we use Trimedia processors hosted by a PC. This paper describes the relationship between the algorithms used for human activity detection and the architectures required to perform these tasks in real time.

Object detection and activity recognition in digital image and video libraries

Abstract: OBJECT DETECTION AND ACTIVITY RECOGNITION IN DIGITAL IMAGE AND VIDEO LIBRARIES by Ibrahim Burak Ozer This thesis is a comprehensive study of object-based image and video retrieval, specifically for car and human detection and activity recognition purposes. The thesis focuses on the problem of connecting low level features to high level semantics by developing relational object and activity presentations. With the rapid growth of multimedia information in forms of digital image and video libraries, there is an increasing need for intelligent database management tools. The traditional text based query systems based on manual annotation process are impractical for today's large libraries requiring an efficient information retrieval system. For this purpose, a hierarchical information retrieval system is proposed where shape, color and motion characteristics of objects of interest are captured in compressed and uncompressed domains. The proposed retrieval method provides object detection and activity recognition at different resolution levels from low complexity to low false rates. The thesis first examines extraction of low level features from images and videos using intensity, color and motion of pixels and blocks. Local consistency based on these features and geometrical characteristics of the regions is used to group object parts. The problem of managing the segmentation process is solved by a new approach that uses object based knowledge in order to group the regions according to a global consistency. A new model-based segmentation algorithm is introduced that uses a feedback from relational representation of the object. The selected unary and binary attributes are further extended for application specific algorithms. Object detection is achieved by matching the relational graphs of objects with the reference model. The major advantages of the algorithm can be summarized as improving the object extraction by reducing the dependence on the low level segmentation process and combining the boundary and region properties. The thesis then addresses the problem of object detection and activity recognition in compressed domain in order to reduce computational complexity. New algorithms for object detection and activity recognition in JPEG images and MPEG videos are developed. It is shown that significant information can be obtained from the compressed domain in order to connect to high level semantics. Since our aim is to retrieve information from images and videos compressed using standard algorithms such as JPEG and MPEG, our approach differentiates from previous compressed domain object detection techniques where the compression algorithms are governed by characteristics of object of interest to be retrieved. An algorithm is developed using the principal component analysis of MPEG motion vectors to detect the human activities; namely, walking, running, and kicking. Object detection in JPEG compressed still images and MPEG I frames is achieved by using DC-DCT coefficients of the luminance and chrominance values in the graph based object detection algorithm. The thesis finally addresses the problem of object detection in lower resolution and monochrome images. Specifically, it is demonstrated that the structural information of human silhouettes can be captured from AC-DCT coefficients. OBJECT DETECTION AND ACTIVITY RECOGNITION IN DIGITAL IMAGE AND VIDEO LIBRARIES by Ibrahim Burak Ozer A Dissertation Submitted to the Faculty of New Jersey Institute of Technology in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy in Electrical Engineering Department of Electrical and Computer Engineering