Adaptive Margin Diversity Regularizer for Handling Data Imbalance in Zero-Shot SBIR

Dutta, T and Singh, A and Biswas, S (2020) Adaptive Margin Diversity Regularizer for Handling Data Imbalance in Zero-Shot SBIR. In: Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), 23 - 28 August 2020, Glasgow, pp. 349-364.

PDF eccv_2020.pdf - Published Version Restricted to Registered users only Download (1MB) | Request a copy

Abstract

Data from new categories are continuously being discovered, which has sparked significant amount of research in developing approaches which generalize to previously unseen categories, i.e. zero-shot setting. Zero-shot sketch-based image retrieval (ZS-SBIR) is one such problem in the context of cross-domain retrieval, which has received lot of attention due to its various real-life applications. Since most real-world training data have a fair amount of imbalance; in this work, for the first time in literature, we extensively study the effect of training data imbalance on the generalization to unseen categories, with ZS-SBIR as the application area. We evaluate several state-of-the-art data imbalance mitigating techniques and analyze their results. Furthermore, we propose a novel framework AMDReg (Adaptive Margin Diversity Regularizer), which ensures that the embeddings of the sketches and images in the latent space are not only semantically meaningful, but they are also separated according to their class-representations in the training set. The proposed approach is model-independent, and it can be incorporated seamlessly with several state-of-the-art ZS-SBIR methods to improve their performance under imbalanced condition. Extensive experiments and analysis justify the effectiveness of the proposed AMDReg for mitigating the effect of data imbalance for generalization to unseen classes in ZS-SBIR. © 2020, Springer Nature Switzerland AG.

Item Type:	Conference Paper
Publication:	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Publisher:	Springer Science and Business Media Deutschland GmbH
Additional Information:	The copyright for this article belongs to Springer Science and Business Media Deutschland GmbH.
Keywords:	Computer vision, Application area; Data imbalance; Model independent; Real-life applications; Sketch-based image retrievals; State of the art; Training data; Training sets, Data handling
Department/Centre:	Division of Electrical Sciences > Electrical Engineering
Date Deposited:	07 Feb 2023 09:29
Last Modified:	07 Feb 2023 09:29
URI:	https://eprints.iisc.ac.in/id/eprint/79998

Actions (login required)

View Item