NFM dataset

该数据集包含了从17段街景视频中提取的18,088个视频帧中的581,108个人脸标注，旨在用于估算戴口罩的比例。数据集包含了针对人脸的戴口罩、未戴口罩以及未知状态的标注，并已对标注的准确性进行了验证。该数据集规模宏大，含有581,108个标注，其任务在于估算戴口罩的比例。

This dataset contains 581,108 facial annotations extracted from 18,088 video frames sourced from 17 street-view videos, and is intended for mask-wearing rate estimation. The dataset includes annotations for three facial states: mask-wearing, non-mask-wearing, and unknown status, and the accuracy of the annotations has been verified. This is a large-scale dataset with 581,108 annotations, whose core task is mask-wearing rate estimation.