HELPING THE SYSTEM
Although precision is usually high when searching on textual annotations, the recall may be disappointing. In this example, we found only four images. Using content-based image retrieval techniques, we should be able to improve the recall (but precision will drop).
To increase the recall, we thus switch to searching on content, and tell the system which images we liked from the initial query by marking them positive (green URLs in the screenshot).
| Prev | Next |
Home: DEMO - The Mirror MMDBMS architecture
Last updated: December 9, 1999
Maintained by: arjen@acm.org