Style-guided zero-shot sketch-based image retrieval

Dutta, T and Biswas, S (2020) Style-guided zero-shot sketch-based image retrieval. In: 30th British Machine Vision Conference 2019, BMVC 2019, 8 - 12 September 2019, Cardiff.

PDF BMVC_2019.pdf - Published Version Restricted to Registered users only Download (699kB) | Request a copy

Abstract

Given a sketch query from a previously unseen category, the goal of zero-shot sketch-based image retrieval (ZS-SBIR) is to retrieve semantically meaningful images from a given database. The knowledge-gap between the seen and unseen categories along with sketch-image domain shift makes this an extremely challenging problem. In this work, we propose a novel framework which decomposes each image and sketch into its domain-independent content and a domain, as well as data-dependent variation/style component. Specifically, given a query sketch and a search set of images, we utilize the image specific styles to guide the generation of fake images using the query content to be used for retrieval. Extensive experiments on two large-scale sketch-image datasets, Sketchy extended and TU-Berlin show that the proposed approach performs better or comparable to the state-of-the-art in both ZS-SBIR and generalized ZS-SBIR protocols. © 2019. The copyright of this document resides with its authors.

Item Type:	Conference Paper
Publication:	30th British Machine Vision Conference 2019, BMVC 2019
Publisher:	BMVA Press
Additional Information:	The copyright for this article belongs to BMVA Press.
Keywords:	Computer vision; Large dataset; Query processing, Data dependent; Domain independents; Image datasets; Image domain; Knowledge gaps; Sketch-based image retrievals; State of the art, Image retrieval
Department/Centre:	Division of Electrical Sciences > Electrical Engineering
Date Deposited:	07 Feb 2023 11:20
Last Modified:	07 Feb 2023 11:20
URI:	https://eprints.iisc.ac.in/id/eprint/80038

Actions (login required)

View Item