Leveraging Image based Prior for Visual Place Recognition

TL;DR

This paper introduces a new scene descriptor for visual place recognition that leverages a library of raw images from sources like Google StreetView and Flickr, enabling cost-effective and readily available landmark matching without requiring spatial information.

Contribution

The paper presents a novel scene descriptor based on raw image libraries, differing from traditional bag-of-words methods, and demonstrates its effectiveness in visual place recognition tasks.

Findings

Outperforms previous scene description methods in recognition accuracy

Uses publicly available image collections, reducing data collection costs

Provides a compact and discriminative scene representation

Abstract

In this study, we propose a novel scene descriptor for visual place recognition. Unlike popular bag-of-words scene descriptors which rely on a library of vector quantized visual features, our proposed descriptor is based on a library of raw image data, such as publicly available photo collections from Google StreetView and Flickr. The library images need not to be associated with spatial information regarding the viewpoint and orientation of the scene. As a result, these images are cheaper than the database images; in addition, they are readily available. Our proposed descriptor directly mines the image library to discover landmarks (i.e., image patches) that suitably match an input query/database image. The discovered landmarks are then compactly described by their pose and shape (i.e., library image ID, bounding boxes) and used as a compact discriminative scene descriptor for the input image. We evaluate the effectiveness of our scene description framework by comparing its performance to that of previous approaches.