Gates Building, Room 100, 3:15-4:30
21 Feb, 1997
Prof. Neil C. Rowe, Naval Postgraduate School, Monterey, CA 93943

Precise and Efficient Access to Captioned Picture Libraries: the MARIE Project

Seminar abstract

We describe the MARIE system for knowledge-based information retrieval of captioned images of the kind found in picture libraries and on the Internet. MARIE exploits the idea that images are much easier to understand with context, especially descriptive text near them, but it also does image analysis. MARIE has five parts
  1. Find the images and captions
  2. Parse and interpret the captions
  3. Segment the images into regions of homogeneous characteristics and classify them
  4. Correlate caption interpretation with image interpretation
  5. Optimize query execution at runtime
MARIE emphasizes domain-independent methods for portability at the expense of some performance, although some domain specification is still required. Experiments show MARIE prototypes run reasonably fast. We illustrate their processing in detail on a single Web page.