Columbia MVSO Image Sentiment Dataset
收藏资源简介:
哥伦比亚MVSO图像情感数据集是由哥伦比亚大学创建的,专注于通过图像表达的情感和情绪。该数据集包含从Flickr网站爬取的700多万张图像,这些图像通过形容词-名词对(ANP)进行标记,旨在通过这些ANP来预测图像的情感倾向。数据集的创建过程中,使用了亚马逊Mechanical Turk平台进行人工标注,确保了数据的质量和多样性。该数据集主要用于评估自动预测图像情感的系统,为情感分析领域提供了重要的基准数据。
The Columbia MVSO Image Emotion Dataset was developed by Columbia University, focusing on emotions and sentiments conveyed through images. This dataset contains over 7 million images crawled from the Flickr website, with each image annotated using Adjective-Noun Pairs (ANPs) to predict the emotional tendencies of images via these ANPs. During the dataset's construction process, manual annotation was carried out via the Amazon Mechanical Turk platform to ensure the quality and diversity of the data. This dataset is primarily used to evaluate automated systems for image emotion prediction, providing critical benchmark data for the field of sentiment analysis.



