3D scene modeling for activity detection

Yunqian Ma, Mike Bazakos, Zheng Wang, Wing Au

Research output: Chapter in Book/Report/Conference proceedingConference contribution

1 Scopus citations


Current computer vision algorithms can process video sequences and perform key low-level functions, such as motion detection, motion tracking, and object classification. This motivates activity detection (e.g. recognizing people's behavior and intent), which is becoming increasingly important. However, they all have severe performance limitations when used over an extended range of applications. They suffer from high false detection rates and missing detection rates, or loss of track due to partial occlusions, etc. Also, activity detection is limited to 2D image domain and is confined to qualitative activities (such as a car entering a region of interest). Adding 3D information will increase the performance of all computer vision algorithms and the activity detection system. In this paper, we propose a unique approach which creates a 3D site model via sensor fusion of laser range finder and a single camera, which then can convert the symbolic features (pixel based) of each object to physical features (e.g. feet or yards). We present experimental results to demonstrate our 3D site model.

Original languageEnglish (US)
Title of host publicationPerspectives in Conceptual Modeling - ER 2005 Workshops CAOIS, BP-UML, CoMoGIS, eCOMO, and QoIS, Proceedings
Number of pages10
StatePublished - Dec 1 2005
Externally publishedYes
EventER 2005 Workshops CAOIS, BP-UML, CoMoGIS, eCOMO, and QoIS - Perspectives in Conceptual Modeling - Klagenfurt, Austria
Duration: Oct 24 2005Oct 28 2005

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume3770 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349


OtherER 2005 Workshops CAOIS, BP-UML, CoMoGIS, eCOMO, and QoIS - Perspectives in Conceptual Modeling


Dive into the research topics of '3D scene modeling for activity detection'. Together they form a unique fingerprint.

Cite this