Transcript Slides with notes about Image Retrieval Systems exercise

What do you understand about how each
system works to index-retrieve images?
Manually Index
Expensive but effective
What do you understand about how each
system works to index-retrieve images?
Visual Similarity (QBIC / Blobworld)
User makes a visual query (i.e. “I want
image(s) that look like this”) – where “look
like” is defined mathematically by the system
(e.g. colour histograms, texture, shape)
What do you understand about how each
system works to index-retrieve images?
Use keywords in HTML (search engines)
?How to select the keywords?
Take from the filename / directory names
Take them from the ALT tag
??Maybe take them from the rest of the text on
the webpage??
What do you think are the strengths and weaknesses
of each approach / system? What kinds of
information needs / queries are they good for?
Manual indexing – good when there’s a business
model
Visual Similarity – match patterns like trademarks,
find a painting whose name you’ve forgotten, find
images to match a colour schemes
Keywords in HTML – best we’ve got for general
web searches