Please use this identifier to cite or link to this item:
Scopus Web of Science® Altmetric
Type: Conference paper
Title: Convolutional Nets Versus Vision Transformers for Diabetic Foot Ulcer Classification
Author: Galdran, A.
Carneiro, G.
Ballester, M.A.G.
Citation: Lecture Notes in Artificial Intelligence, 2022 / Yap, M.H., Cassidy, B., Kendrick, C. (ed./s), vol.13183 LNCS, pp.21-29
Publisher: Springer International Publishing
Publisher Place: Cham, Switzerland
Issue Date: 2022
Series/Report no.: Lecture Notes in Computer Science; 13183
ISBN: 9783030949068
ISSN: 0302-9743
Conference Name: Diabetic Foot Ulcers Grand Challenge (DFUC) (27 Sep 2021 : Strasbourg, France)
Editor: Yap, M.H.
Cassidy, B.
Kendrick, C.
Statement of
Adrian Galdran, Gustavo Carneiro and Miguel A. González Ballester
Abstract: This paper compares well-established Convolutional Neural Networks (CNNs) to recently introduced Vision Transformers for the task of Diabetic Foot Ulcer Classification, in the context of the DFUC 2021 Grand-Challenge, in which this work attained the first position. Comprehensive experiments demonstrate that modern CNNs are still capable of outperforming Transformers in a low-data regime, likely owing to their ability for better exploiting spatial correlations. In addition, we empirically demonstrate that the recent Sharpness-Aware Minimization (SAM) optimization algorithm improves considerably the generalization capability of both kinds of models. Our results demonstrate that for this task, the combination of CNNs and the SAM optimization process results in superior performance than any other of the considered approaches.
Keywords: Diabetic Foot Ulcer Classification; Vision Transformers; Convolutional Neural Networks; Sharpness-Aware Optimization
Description: Conference was held in Conjunction with MICCAI 2021.
Rights: © 2022 Springer Nature Switzerland AG
DOI: 10.1007/978-3-030-94907-5_2
Grant ID:
Published version:
Appears in Collections:Computer Science publications

Files in This Item:
There are no files associated with this item.

Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.