FER+

FER代表人脸表情识别Plus，它是原始FER数据集的扩展。
该数据集为原始数据集提供了一组新的标签，每张图片都由10个众包注释，这不仅为面部表情识别提供了更好的地面真相，还使研究人员能够更准确地估计面部情绪的概率分布。这些图像被重新分类为八种情感类型: 中立，快乐，惊讶，悲伤，愤怒，厌恶，恐惧和蔑视。

FER, short for Facial Expression Recognition Plus, is an extension of the original FER dataset. It provides a new set of labels for the original dataset, where each image is annotated by 10 crowd-sourced annotators. This not only delivers more reliable ground truth for facial expression recognition tasks, but also enables researchers to accurately estimate the probability distribution of facial emotions. All images in this dataset are reclassified into eight emotional categories: Neutral, Happiness, Surprise, Sadness, Anger, Disgust, Fear, and Contempt.

AI搜集汇总

背景与挑战

背景概述

FER+是原始FER数据集的扩展版本，通过众包注释为每张图片提供新标签，增强了面部表情识别的地面真相和概率分布估计。该数据集重新分类了八种情感类型，包括中立、快乐、惊讶等，由微软于2016年发布，适用于面部情绪识别研究。