LCAF
收藏资源简介:
LCAF数据集是由复旦大学收集并发布的一个大规模跨年龄人脸数据集,包含170万张人脸图像,每张图像都标注了年龄和性别信息。该数据集旨在推动年龄不变人脸识别(AIFR)和人脸年龄合成(FAS)的研究。数据集的创建过程涉及使用公共的Azure面部API对来自MS-Celeb-1M数据集的面部进行年龄和性别估计,随后通过随机抽样和人工校正确保标注的准确性。LCAF数据集不仅适用于AIFR和FAS的研究,还可用于其他面部相关的研究任务,如面部年龄估计的预训练。此外,为了促进追踪长期失踪儿童的应用,该数据集还构建了一个新的基准,包含相同身份的儿童和成人面部图像,专门设计用于跨年龄人脸识别的未来评估。
The LCAF dataset is a large-scale cross-age face dataset collected and released by Fudan University, comprising 1.7 million face images, each annotated with age and gender information. This dataset aims to advance research in Age-Invariant Face Recognition (AIFR) and Face Age Synthesis (FAS). The construction of the dataset involved using the public Azure Face API to estimate age and gender for faces from the MS-Celeb-1M dataset, followed by random sampling and manual verification to ensure annotation accuracy. The LCAF dataset is not only applicable to AIFR and FAS research, but also can be used for other facial-related research tasks, such as pre-training for face age estimation. Furthermore, to facilitate applications related to tracking long-term missing children, the dataset has also established a new benchmark containing child and adult face images of identical identities, specifically designed for future evaluations of cross-age face recognition.

- 1When Age-Invariant Face Recognition Meets Face Age Synthesis: A Multi-Task Learning Framework and A New Benchmark复旦大学 · 2022年



