King's College London

TY - JOUR

T1 - Discovering Regression-Detection Bi-knowledge Transfer for Unsupervised Cross-Domain Crowd Counting

AU - Liu, Yuting

AU - Wang, Zheng

AU - Shi, Miaojing

AU - Satoh, Shin'ichi

AU - Zhao, Qijun

AU - Yang, Hongyu

N1 - Funding Information: The research was partly supported by National Natural Science Foundation of China (No. 62176170, 61828602, 62066042, 61971005), JST CREST Grant (JPMJCR1686), and Grant-in-Aid for JSPS Fellows (18F18378). Publisher Copyright: © 2022

PY - 2022/7/14

Y1 - 2022/7/14

N2 - Despite impressive progress in crowd counting over the last years, it is still an open challenge to reliably count crowds across visual domains. This paper addresses this setting, presenting an unsupervised cross-domain crowd counting framework able to perform unsupervised adaptation across domains with available unlabeled target data. We achieve this by learning to discover bi-knowledge transfer between regression- and detection-based models from a labeled source domain. The dual source knowledge of the two models is heterogeneous and complementary as they capture different modalities of crowd distribution. Specifically, we start by formulating the mutual transformations between the outputs of regression- and detection-based models as two scene-agnostic transformers which enable knowledge transfer between the two models. Given the regression- and detection-based models and their mutual transformers learnt on the source, we then introduce a self-supervised co-training scheme to encourage the knowledge transfer between the two models on the target. We further enhance the model adaptation with our modified mixup augmentation strategy. A thorough benchmark analysis against the most recent cross-domain crowd counting methods and detailed ablation studies show the advantage of our method.

AB - Despite impressive progress in crowd counting over the last years, it is still an open challenge to reliably count crowds across visual domains. This paper addresses this setting, presenting an unsupervised cross-domain crowd counting framework able to perform unsupervised adaptation across domains with available unlabeled target data. We achieve this by learning to discover bi-knowledge transfer between regression- and detection-based models from a labeled source domain. The dual source knowledge of the two models is heterogeneous and complementary as they capture different modalities of crowd distribution. Specifically, we start by formulating the mutual transformations between the outputs of regression- and detection-based models as two scene-agnostic transformers which enable knowledge transfer between the two models. Given the regression- and detection-based models and their mutual transformers learnt on the source, we then introduce a self-supervised co-training scheme to encourage the knowledge transfer between the two models on the target. We further enhance the model adaptation with our modified mixup augmentation strategy. A thorough benchmark analysis against the most recent cross-domain crowd counting methods and detailed ablation studies show the advantage of our method.

UR - http://www.scopus.com/inward/record.url?scp=85129607306&partnerID=8YFLogxK

U2 - 10.1016/j.neucom.2022.04.107

DO - 10.1016/j.neucom.2022.04.107

M3 - Article

VL - 494

SP - 418

EP - 431

JO - NEUROCOMPUTING

JF - NEUROCOMPUTING

SN - 0925-2312

ER -