 |
City name estimation by learning large image dataset using random forest |
|
In recent years, by the spread of sharing services of images and videos,
we have come to be able to acquire images and photographs of several cities
in the world from the Internet. This in- formation can expect to apply to the
three-dimensional model generation of the city and the frequent update of the
map or scene simulation, but may cause the presentation of the wrong information
when the data of various cities mingles. Therefore, we aim for the global
localization of images which are taken at various cities in the world. In the method, we introduce random forest to learn the information of each city from a street view image, then estimate the city where an inputted image taken from.
Figure1. Evaluation of the conviction degree
Figure2. Diagrammatic view of the local estimated technique
Publications
|
| Kawasaki Laboratory |