面向高清人体图像生成的数据基准与模型框架  

Data benchmark and model framework for high-definition human image generation

作　　者：徐正国 普碧才 秦建明 项炎平 彭振江 宋纯锋 Xu Zhengguo;Pu Bicai;Qin Jianming;Xiang Yanping;Peng Zhenjiang;Song Chunfeng(Nujiang Power Supply Bureau,Yunnan Power Grid Co.,Ltd.,Nujiang 673199,China;Intelligent Interconnection Technology Co.,Beijing 100181,China;Center for Research on Intelligent Perception and Computing,Institute of Automation,Chinese Academy of Sciences,Beijing 100190,China)

机构地区：[1]云南电网有限责任公司怒江供电局,怒江673199 [2]智慧互通科技股份有限公司,北京100081 [3]中国科学院自动化研究所智能感知与计算研究中心,北京100190

出　　处：《中国图象图形学报》2025年第2期375-390,共16页Journal of Image and Graphics

基　　金：国家自然科学基金项目(62006231)。

摘　　要：目的姿态引导下的人物图像生成具有广泛的应用潜力,受到了广泛关注。低分辨率场景的姿态引导人物图像生成任务取得了很大成功。然而在高分辨率场景下,现有的人体姿态迁移数据集存在分辨率低或多样性差等问题,同时也缺乏相关高分辨率图像生成方法。针对这一问题,构建了具有多模态辅助数据的大规模高清人物图像数据集PersonHD。方法PersonHD数据集收集了包含100个不同人物的299817幅图像。在提出的PersonHD基础上,基于现有数据集的公共设置,本文进一步构建了两个不同分辨率下的评测基准,并设计了一个实用的高分辨率人物图像生成框架,为评估最先进的姿态引导人物图像生成方法提供了一个新的平台。结果与现有数据集相比,PersonHD在更高的图像分辨率、更多样化的人物姿态和更大规模的样本方面具有显著的优势。基于PersonHD数据集,实验在两个不同分辨率的评测基准上系统地评估了当前具有代表性的姿态引导人物图像生成方法,并对本文提出框架各模块的有效性进行了系统验证。实验结果表明,该框架具有良好的效果。结论本文提出的高清人物图像生成基准数据集具有高分辨率数据规模大、多样性强等特点,有助于更为全面地评估姿态引导下的人物图像生成算法。本文的数据集和代码可在https://github.com/BraveGroup/PersonHD上获得。Objective Pose-guided person image generation has attracted considerable attention because of its wide applica⁃tion potential.In the early stages of development,researchers relied mainly on manually designing features and models,matching key points between different characters,and then achieving pose transfer via interpolation or transformation.With the rapid development of deep learning technology,the emergence of generative adversarial networks(GANs)has led to considerable progress in posture transfer.GANs can learn and generate realistic images,and variants of related genera⁃tive adversarial networks have been widely used in pose transfer tasks.Moreover,deep learning has made progress in key point detection technology.Advanced key point detection models,such as OpenPose,can more accurately capture human pose information,providing tremendous assistance for the development of algorithms in related fields and the construction of datasets.Recent works have achieved great success in pose-guided person image generation tasks with low-definition scenes.However,in high-resolution scenes,existing human pose transfer datasets suffer from low resolution or poor diver⁃sity,and relevant high-resolution image generation methods are lacking.This issue is addressed by constructing a largescale high-definition human image dataset named PersonHD with multimodal auxiliary data.Method This study constructs a large-scale,high-resolution human image dataset called PersonHD.Compared with other datasets,this dataset has sev⁃eral advantages.1)Higher image resolution:the cropped human images in PersonHD have a resolution of 1520×880 pix⁃els.2)More diverse pose variations:the actions of the subjects are closer to real-life scenarios,introducing more finegrained nonrigid deformation of the human body.3)Larger image size.The PersonHD dataset contains 299817 images from 100 different people in 4000 videos.On the basis of the proposed PersonHD,this study further constructs two bench⁃marks and designs a practical high-resolution

关 键 词：人物图像合成 姿态引导迁移 高清数据集 低分辨率基准 高分辨率基准 

分 类 号：TP392[自动化与计算机技术—计算机应用技术]