LAMP-HQ
收藏资源简介:
LAMP-HQ是一个大规模、多姿态、高质量的近红外-可见光人脸识别数据库,由中国科学院自动化研究所创建。该数据库包含56,788张近红外图像和16,828张可见光图像,涵盖573个主题,具有多种姿态、光照、属性、场景和配件的广泛多样性。创建过程涉及使用Canon-7D和AuthenMetric-CE31S相机在五种不同的光照场景下采集图像,并通过手动检查去除模糊或不完整的图像。LAMP-HQ数据库旨在推动近红外-可见光人脸分析的研究,特别是在解决跨光谱匹配难题方面。
LAMP-HQ is a large-scale, multi-pose, high-quality near-infrared (NIR)-visible light face recognition database created by the Institute of Automation, Chinese Academy of Sciences. This database contains 56,788 near-infrared images and 16,828 visible light images, covering 573 subjects, and features extensive diversity across poses, illumination, attributes, scenes and accessories. Its creation involved image acquisition using Canon-7D and AuthenMetric-CE31S cameras under five different illumination scenarios, followed by manual inspection to remove blurry or incomplete images. The LAMP-HQ database aims to advance research on near-infrared-visible light facial analysis, particularly in addressing the cross-spectrum face matching challenge.




