DFEW

该数据集包含了来自不同主题电影的超过16,000个视频片段，旨在用于视频中的人脸表情识别任务。这些片段涵盖了多样化的内容，有助于提升表情识别算法的准确性和泛化能力。

This dataset comprises over 16,000 video clips sourced from films with diverse themes, and is specifically designed for the task of facial expression recognition in videos. The diverse content covered by these clips helps improve the accuracy and generalization capability of facial expression recognition algorithms.

搜集汇总

背景与挑战

背景概述

DFEW是一个包含16372个电影视频片段的大规模动态面部表情数据库，具有极端光照/遮挡等挑战性场景，提供7维表情分布和离散情绪标注，所有片段均经过专家多轮独立标注。