Location Recognition

Research Project Goal

There are several resources of large databases of city views. We would like to explore the use of the imagery possible aid for navigation, tourism, search for interesting landmarks and points of interest.

Research Scope

You are given a large database of images of city's landmarks/buildings. Given a novel query view recognize which landmark/building is in the view.

Sample Data

Two options:

• Image database of about 200 buildings of city of Zurich. Traning and testing sets are available at www.vision.ee.ethz.ch/showroom/zubud/index.en.html
• Collect your own dataset of all the buildings at Stanford University from different viewpoints, different times of day (different lighting conditions) at least 3-5 images per building. Use the images of all the buildings on the web as query views (or split the database you collected into training and testing set).

Tasks

• extract suitable features (points, lines, SIFT fetaures, color features, wavelet coefficients)
• propose an efficient indexing and classification scheme for finding the closest view in the database
• Optional: suggest/propose a cell phone interface

Challenges

• choice suitable image features suitable for to disambiguating all the classes, effiency of the classification is also a concern here

Research Project Status

student names here

Point of Contact

Jana Kosecka

Midterm Report

not yet submitted

Final Report

not yet submitted