Mobile Multi-View Object Image Search

TL;DR

This paper presents a mobile multi-view object image search system that captures multiple object views to improve retrieval accuracy, utilizing a client-server architecture with fusion methods for enhanced performance.

Contribution

It introduces a novel mobile multi-view search system and analyzes fusion strategies, demonstrating improved accuracy over single-view methods.

Findings

Multi-view search significantly outperforms single-view search in accuracy.

Fusion methods impact the effectiveness of multi-view object retrieval.

Comprehensive analysis on existing and new databases validates the approach.

Abstract

High user interaction capability of mobile devices can help improve the accuracy of mobile visual search systems. At query time, it is possible to capture multiple views of an object from different viewing angles and at different scales with the mobile device camera to obtain richer information about the object compared to a single view and hence return more accurate results. Motivated by this, we developed a mobile multi-view object image search system, using a client-server architecture. Multi-view images of objects acquired by the mobile clients are processed and local features are sent to the server, which combines the query image representations with early/late fusion methods based on bag-of-visual-words and sends back the query results. We performed a comprehensive analysis of early and late fusion approaches using various similarity functions, on an existing single view and a new multi-view object image database. The experimental results show that multi-view search provides significantly better retrieval accuracy compared to single view search.