AssocFormer: Association Transformer for Multi-label Classification

Xin Xing; Chong Peng; Yu Zhang; Ai Ling Lin; Nathan Jacobs

AssocFormer: Association Transformer for Multi-label Classification

Xin Xing, Chong Peng, Yu Zhang, Ai Ling Lin, Nathan Jacobs

Research output: Contribution to conference › Paper › peer-review

Abstract

The goal of multi-label image classification is to predict a set of labels for a single image. Recent work has shown that explicitly modeling the co-occurrence relationship between classes is critical for achieving good performance on this task. State-of-the-art approaches model this using graph convolutional networks, which are complex and computationally expensive. We propose a novel, efficient association module as an alternative. This is coupled with a transformer-based feature-extraction backbone. The proposed model was evaluated using two standard datasets: MS-COCO and PASCAL VOC. The results show that the proposed model outperforms several strong baseline models.

Original language	English
State	Published - 2022
Event	33rd British Machine Vision Conference Proceedings, BMVC 2022 - London, United Kingdom Duration: Nov 21 2022 → Nov 24 2022

Conference

Conference	33rd British Machine Vision Conference Proceedings, BMVC 2022
Country/Territory	United Kingdom
City	London
Period	11/21/22 → 11/24/22

Cite this

@conference{aef27ea2ad2043cda10d8e2bb9b8de0f,

title = "AssocFormer: Association Transformer for Multi-label Classification",

abstract = "The goal of multi-label image classification is to predict a set of labels for a single image. Recent work has shown that explicitly modeling the co-occurrence relationship between classes is critical for achieving good performance on this task. State-of-the-art approaches model this using graph convolutional networks, which are complex and computationally expensive. We propose a novel, efficient association module as an alternative. This is coupled with a transformer-based feature-extraction backbone. The proposed model was evaluated using two standard datasets: MS-COCO and PASCAL VOC. The results show that the proposed model outperforms several strong baseline models.",

author = "Xin Xing and Chong Peng and Yu Zhang and Lin, {Ai Ling} and Nathan Jacobs",

note = "Publisher Copyright: {\textcopyright} 2022. The copyright of this document resides with its authors. It may be distributed unchanged freely in print or electronic forms.; 33rd British Machine Vision Conference Proceedings, BMVC 2022 ; Conference date: 21-11-2022 Through 24-11-2022",

year = "2022",

language = "English",

}

TY - CONF

T1 - AssocFormer

T2 - 33rd British Machine Vision Conference Proceedings, BMVC 2022

AU - Xing, Xin

AU - Peng, Chong

AU - Zhang, Yu

AU - Lin, Ai Ling

AU - Jacobs, Nathan

PY - 2022

Y1 - 2022

N2 - The goal of multi-label image classification is to predict a set of labels for a single image. Recent work has shown that explicitly modeling the co-occurrence relationship between classes is critical for achieving good performance on this task. State-of-the-art approaches model this using graph convolutional networks, which are complex and computationally expensive. We propose a novel, efficient association module as an alternative. This is coupled with a transformer-based feature-extraction backbone. The proposed model was evaluated using two standard datasets: MS-COCO and PASCAL VOC. The results show that the proposed model outperforms several strong baseline models.

AB - The goal of multi-label image classification is to predict a set of labels for a single image. Recent work has shown that explicitly modeling the co-occurrence relationship between classes is critical for achieving good performance on this task. State-of-the-art approaches model this using graph convolutional networks, which are complex and computationally expensive. We propose a novel, efficient association module as an alternative. This is coupled with a transformer-based feature-extraction backbone. The proposed model was evaluated using two standard datasets: MS-COCO and PASCAL VOC. The results show that the proposed model outperforms several strong baseline models.

UR - http://www.scopus.com/inward/record.url?scp=85169100888&partnerID=8YFLogxK

M3 - Paper

AN - SCOPUS:85169100888

Y2 - 21 November 2022 through 24 November 2022

ER -

AssocFormer: Association Transformer for Multi-label Classification

Abstract

Conference

Other files and links

Fingerprint

Cite this