Manchester Grand Hyatt Hotel
Manchester Grand Hyatt Hotel
复制标题
DOI：
10.1145/2506364.2506369
复制
发表时间：
2013-10
期刊：
影响因子：
--
通讯作者：
中科院分区：
文献类型：
--
作者：
电脑游戏中的情感互动是一个具有若干新挑战的新领域，例如以稳健的方式检测玩家的面部表情（如快乐、悲伤、惊讶）。在本文中，我们描述了一个众包的努力，用于创建一个大规模的图像数据集的基本真相，这些数据集捕获了玩电脑游戏的用户。电脑游戏的设计是为了引出一个特定的面部表情，游戏将根据检测到的表情给玩家打分。在设计众包任务时，考察的一些变量包括：奖励、标签限制、黄金问题、工人的位置。最后，我们设计了一个大型的标签作业，以最大限度地提高工人的一致性。每张带有面部表情的图片都被贴上了以下表情标签之一：快乐、愤怒、厌恶、蔑视、悲伤、恐惧、惊讶和中性。该数据集包括4万多张图像、工人的判断、游戏检测到的面部表情以及玩家应该表现出的面部表情。
Affective-interaction in computer games is a novel area with several new challenges, such as detecting players' facial expressions (e.g., happy, sad, surprise) in a robust manner. In this paper we describe a crowdsourcing effort for creating the ground-truth of a large-scale dataset of images capturing users playing a computer game. The computer game is designed to elicit a particular facial expressions and the game will score the player according to the detected expression. For designing the crowdsourcing task, some of the examined variables include: reward, tagging limits, golden questions, workers' location. In the end, we designed a large tagging job to maximize workers agreement. Each image with a facial expressions is tagged with one of the following expressions labels: happy, anger, disgust, contempt, sad, fear, surprise, and neutral. The dataset included over 40,000 images, the workers' judgments, the game's detected facial expression and what facial expression the player should be performing.