LMVD

该数据集是一个为野外抑郁检测而建立的大规模多模态视频日志集合，包含了来自1475名参与者在四个多媒体平台上的1823个样本，总时长达到214小时。这些样本中既有抑郁视频日志，也有非抑郁视频日志，并且都对视频质量进行了标注，旨在提升自动抑郁检测研究与应用的质量。该数据集的规模为1823个样本，任务目标是抑郁检测。

This dataset is a large-scale multimodal video log collection developed for in-the-wild depression detection. It encompasses 1823 samples from 1475 participants across four multimedia platforms, with a total duration of 214 hours. These samples include both depressive and non-depressive video logs, and all have been annotated with video quality labels, aiming to improve the quality of automatic depression detection research and practical applications. The dataset comprises 1823 samples, with the targeted task being depression detection.

搜集汇总

背景与挑战

背景概述

LMVD是一个用于抑郁症检测的大规模多模态vlog数据集，包含音频和视觉特征，并提供相关代码和运行指南。