Decoupled Mixup for Out-of-Distribution Visual Recognition

Haozhe Liu, Wentian Zhang, Jinheng Xie, Haoqian Wu, Bing Li*, Ziqi Zhang, Yuexiang Li, Yawen Huang, Bernard Ghanem, Yefeng Zheng

*Corresponding author for this work

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

Convolutional neural networks (CNN) have demonstrated remarkable performance, when the training and testing data are from the same distribution. However, such trained CNN models often largely degrade on testing data which is unseen and Out-Of-the-Distribution (OOD). To address this issue, we propose a novel “Decoupled-Mixup" method to train CNN models for OOD visual recognition. Different from previous work combining pairs of images homogeneously, our method decouples each image into discriminative and noise-prone regions, and then heterogeneously combine these regions of image pairs to train CNN models. Since the observation is that noise-prone regions such as textural and clutter background are adverse to the generalization ability of CNN models during training, we enhance features from discriminative regions and suppress noise-prone ones when combining an image pair. To further improves the generalization ability of trained models, we propose to disentangle discriminative and noise-prone regions in frequency-based and context-based fashions. Experiment results show the high generalization performance of our method on testing data that are composed of unseen contexts, where our method achieves 85.76% top-1 accuracy in Track-1 and 79.92% in Track-2 in NICO Challenge. The source code is available at https://github.com/HaozheLiu-ST/NICOChallenge-OOD-Classification.

Original languageEnglish (US)
Title of host publicationComputer Vision – ECCV 2022 Workshops, Proceedings
EditorsLeonid Karlinsky, Tomer Michaeli, Ko Nishino
PublisherSpringer Science and Business Media Deutschland GmbH
Pages451-464
Number of pages14
ISBN (Print)9783031250743
DOIs
StatePublished - 2023
Event17th European Conference on Computer Vision, ECCV 2022 - Tel Aviv, Israel
Duration: Oct 23 2022Oct 27 2022

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume13806 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference17th European Conference on Computer Vision, ECCV 2022
Country/TerritoryIsrael
CityTel Aviv
Period10/23/2210/27/22

Bibliographical note

Funding Information:
Acknowledgements. We would like to thank the efforts of the NICO Challenge officials, who are committed to maintaining the fairness and openness of the competition. We also could not have undertaken this paper without efforts of every authors. This work was supported in part by the Key-Area Research and Development Program of Guangdong Province, China (No. 2018B010111001), National Key R&D Program of China (2018YFC2000702), in part by the Scientific and Technical Innovation 2030-“New Generation Artificial Intelligence” Project (No. 2020AAA0104100) and in part by the King Abdullah University of Science and Technology (KAUST) Office of Sponsored Research through the Visual Computing Center (VCC) funding.

Funding Information:
We would like to thank the efforts of the NICO Challenge officials, who are committed to maintaining the fairness and openness of the competition. We also could not have undertaken this paper without efforts of every authors. This work was supported in part by the Key-Area Research and Development Program of Guangdong Province, China (No. 2018B010111001), National Key R&D Program of China (2018YFC2000702), in part by the Scientific and Technical Innovation 2030“New Generation Artificial Intelligence” Project (No. 2020AAA0104100) and in part by the King Abdullah University of Science and Technology (KAUST) Office of Sponsored Research through the Visual Computing Center (VCC) funding.

Publisher Copyright:
© 2023, The Author(s), under exclusive license to Springer Nature Switzerland AG.

ASJC Scopus subject areas

  • Theoretical Computer Science
  • General Computer Science

Fingerprint

Dive into the research topics of 'Decoupled Mixup for Out-of-Distribution Visual Recognition'. Together they form a unique fingerprint.

Cite this