EmoSet
收藏资源简介:
EmoSet是由深圳大学创建的一个大规模视觉情感数据集,包含330万张图像,其中118,102张图像经过人工标注。该数据集不仅标注了八种情感类别,还包含了六种情感属性,如亮度、色彩丰富度、场景类型、对象类别、面部表情和人类动作。EmoSet的数据来源多样,包括社交媒体和艺术作品,旨在通过这些丰富的属性帮助理解和预测视觉刺激引起的情感反应。数据集的创建过程涉及从多个来源收集图像,并通过人工和机器学习方法进行标注。EmoSet的应用领域广泛,包括情感计算、图像美学评估和智能广告等,旨在解决如何通过视觉内容理解和预测人类情感的问题。
EmoSet is a large-scale visual emotion dataset created by Shenzhen University, which contains 3.3 million images, among which 118,102 images have been manually annotated. This dataset is not only annotated with eight emotion categories, but also includes six emotional attributes, such as brightness, color richness, scene type, object category, facial expression and human action. EmoSet has diverse data sources, including social media and artworks, aiming to help understand and predict emotional responses evoked by visual stimuli through these rich attributes. The development process of the dataset involves collecting images from multiple sources and annotating them via both manual and machine learning approaches. EmoSet has a wide range of application scenarios, including affective computing, image aesthetic assessment, intelligent advertising and so on, aiming to address the problem of how to understand and predict human emotions through visual content.




