Attentional feature fusion

Yimian Dai, Fabian Gieseke, Stefan Oehmcke, Yiquan Wu, Kobus Barnard

Research output: Chapter in Book/Report/Conference proceedingConference contribution

18 Scopus citations

Abstract

Feature fusion, the combination of features from different layers or branches, is an omnipresent part of modern network architectures. It is often implemented via simple operations, such as summation or concatenation, but this might not be the best choice. In this work, we propose a uniform and general scheme, namely attentional feature fusion, which is applicable for most common scenarios, including feature fusion induced by short and long skip connections as well as within Inception layers. To better fuse features of inconsistent semantics and scales, we propose a multiscale channel attention module, which addresses issues that arise when fusing features given at different scales. We also demonstrate that the initial integration of feature maps can become a bottleneck and that this issue can be alleviated by adding another level of attention, which we refer to as iterative attentional feature fusion. With fewer layers or parameters, our models outperform state-of-the-art networks on both CIFAR-100 and ImageNet datasets, which suggests that more sophisticated attention mechanisms for feature fusion hold great potential to consistently yield better results compared to their direct counterparts. Our codes and trained models are available online1.

Original languageEnglish (US)
Title of host publicationProceedings - 2021 IEEE Winter Conference on Applications of Computer Vision, WACV 2021
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages3559-3568
Number of pages10
ISBN (Electronic)9780738142661
DOIs
StatePublished - Jan 2021
Event2021 IEEE Winter Conference on Applications of Computer Vision, WACV 2021 - Virtual, Online, United States
Duration: Jan 5 2021Jan 9 2021

Publication series

NameProceedings - 2021 IEEE Winter Conference on Applications of Computer Vision, WACV 2021

Conference

Conference2021 IEEE Winter Conference on Applications of Computer Vision, WACV 2021
Country/TerritoryUnited States
CityVirtual, Online
Period1/5/211/9/21

ASJC Scopus subject areas

  • Computer Vision and Pattern Recognition
  • Computer Science Applications

Fingerprint

Dive into the research topics of 'Attentional feature fusion'. Together they form a unique fingerprint.

Cite this