LEARNING 3D OBJECT-CENTRED APPEARANCE MODELS FOR TRACKING

This paper presents a hypothesis verification strategy for 3D object recognition. This methodology integrates 3D object-centered and 2D appearance-based representations in computer vision which leads to improved hypothesis verification. The approach is demonstrated on real-world image sequences from traffic surveillance and compared to edge-based iconic evaluation techniques.

• Supplemental Notes:
  • Publication Date: 1999 IEEE Service Center, Piscataway NJ
• Corporate Authors:

  University of Reading. Dept. of Computer Science

  ,    
• Authors:
  • Ferryman, J
  • Worrall, A
• Conference:
  • IEEE Workshop on the Integration of Appearance and Geometric Methods in Object Recognition (1999 : Ft. Collins, Colo.)
• Publication Date: 1999

Language

• English

Media Info

• Pagination: p. 34-43

Subject/Index Terms

• TRT Terms: Artificial intelligence; Computer vision; Traffic surveillance
• Subject Areas: Data and Information Technology;

Filing Info

• Accession Number: 00780577
• Record Type: Publication
• Source Agency: UC Berkeley Transportation Library
• Files: PATH
• Created Date: Jan 5 2000 12:00AM