While searching for relevant Web pages based on their textual content is among the most common and useful Internet tools available today, searching for relevant images from the Web is still an inconvenient and ineffective process. The most common approach, like that for textual documents, is still based on keyword search. The major limitation of this approach is that it requires prior annotation of the images which is both very tedious and subjective. A more promising approach is to allow the user to simply present a sample image as query. An image-based search engine then makes use of the query image to retrieve similar images from the Web according to some predefined or adaptive similarity criterion. An illustrative example is shown below. The query image is shown with a red frame around it, followed by some retrieved images with their similarity measures decreasing from left to right. Based on the prior research experience of the advisor in this problem, participating students will be guided to make interesting explorations along this direction by building such an image retrieval system.