FaTNET: Feature-alignment transformer network for human pose transfer

doi:10.1016/j.patcog.2025.111626

	FaTNET: Feature-alignment transformer network for human pose transfer
	Luo, Yu 1; Yuan, Chengzhi 1; Gao, Lin 2; Xu, Weiwei 3; Yang, Xiaosong 4; Wang, Pengjie 1
	2025-09-01
发表期刊	PATTERN RECOGNITION
ISSN	0031-3203
卷号	165 页码:10
摘要	Pose-guided person image generation involves converting an image of a person from a source pose pose. This task presents significant challenges due to the extensive variability and occlusion. Existing heavily rely on CNN-based architectures, which are constrained by their local receptive fields and often to preserve the details of style and shape. To address this problem, we propose a novel framework pose transfer with transformers, which can employ global dependencies and keep local features as proposed framework consists of transformer encoder, feature alignment network and transformer network, enabling the generation of realistic person images with desired poses. The core idea of our is to obtain a novel prior image aligned with the target image through the feature alignment network embedded and disentangled feature space, and then synthesize the final fine image through the transformer synthetic network by recurrently warping the result of previous stage with the correlation matrix aligned features and source images. In contrast to previous convolution and non-local methods, employ the global receptive field and preserve detail features as well. The results of qualitative and quantitative experiments demonstrate the superiority of our model in human pose transfer.
关键词	People image generation Human pose transfer Generative adversarial network Transformers
DOI	10.1016/j.patcog.2025.111626
收录类别	SCI
语种	英语
资助项目	European Union[900025] ; Natural Science Foundation of Liaoning[2023-MS-133] ; Liaoning Provincial Education Department, China[LJ242412026018] ; Marie Curie Actions (MSCA)[900025]
WOS研究方向	Computer Science ; Engineering
WOS类目	Computer Science, Artificial Intelligence ; Engineering, Electrical & Electronic
WOS记录号	WOS:001466546600001
出版者	ELSEVIER SCI LTD
引用统计
文献类型	期刊论文
条目标识符	http://119.78.100.204/handle/2XEOYT63/40591
专题	中国科学院计算技术研究所期刊论文_英文
通讯作者	Wang, Pengjie
作者单位	1.Dalian Minzu Univ, 18 Liaohe West Rd, Dalian 116000, Peoples R China 2.Chinese Acad Sci, Inst Comp Technol, 6 Kexueyuan South Rd, Beijing 100080, Peoples R China 3.Zhejiang Univ, 866 Yuhangtang Rd, Hangzhou 310058, Peoples R China 4.Bournemouth Univ, Poole BH12 5BB, Dorset, England
推荐引用方式 GB/T 7714	Luo, Yu,Yuan, Chengzhi,Gao, Lin,et al. FaTNET: Feature-alignment transformer network for human pose transfer[J]. PATTERN RECOGNITION,2025,165:10.
APA	Luo, Yu,Yuan, Chengzhi,Gao, Lin,Xu, Weiwei,Yang, Xiaosong,&Wang, Pengjie.(2025).FaTNET: Feature-alignment transformer network for human pose transfer.PATTERN RECOGNITION,165,10.
MLA	Luo, Yu,et al."FaTNET: Feature-alignment transformer network for human pose transfer".PATTERN RECOGNITION 165(2025):10.