Labeling Spain with Stanford

We present an end-to-end framework for outdoor scene region decomposition, learned on a small set of randomly selected images, that generalizes well to multiple datasets containing images from around the world. We discuss the different aspects of the framework especially a generalized variational in...

Full description

Saved in:
Bibliographic Details
Published in:IEEE transactions on image processing 2013-10
Main Authors: Zhou, Yingbo, Nwogu, Ifeoma, Govindaraju, Venu
Format: Article
Language:English
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:We present an end-to-end framework for outdoor scene region decomposition, learned on a small set of randomly selected images, that generalizes well to multiple datasets containing images from around the world. We discuss the different aspects of the framework especially a generalized variational inference method with better approximations to the true marginals of a graphical model. Experimentally, we explain why the framework is robust and performs competitively on many diverse scene datasets, including several unseen scene types. We have obtained high pixel-level accuracies (80%) in three of the four datasets, which include a benchmark dataset known as the Stanford BG dataset. Our model obtained over 70% accuracy on the fourth dataset, which contained a number of indoor and close-up images that are significantly different from our training examples.
ISSN:1941-0042