Image Sentiment Analysis (2016)

Image Sentiment Analysis Using Latent CorrelationsAmong Visual, Textual, and Sentiment Views

Abstract: As many users routinely share images from their daily lives on the Internet, analyzing the sentiments embedded in such content can facilitate the extraction of events and emerging trends. In this study, we construct an image sentiment classifier by integrating visual features, textual information accompanying the posts, and external sentiment lexicons. To support our experiments, we created a dataset consisting of images and sentiment labels through a crowdsourcing process.

Authors: Marie Katsurai, Shin’ichi Satoh

Publication venue: ICASSP2016

Dataset

To evaluate the performance of image sentiment classification, we collected a set of images from Flickr and Instagram, and then prepared their sentiment labels via crowdsourcing. For each image, three workers were asked to provide a sentiment score. They could choose on a discrete five-point scale labeled with “highly positive,” “positive,” “neutral,” “negative,” and “highly negative.” The datasets with sentiment labels (the number of users for each sentiment polarity) are available. that we divided the whole dataset into three batches for download.

NOTE: Our data collection was conducted using the Instagram API prior to changes in Instagram’s data access policies.

• Flickr dataset file 1
• Flickr dataset file 2
• Flickr dataset file 3
• Instagram dataset file 1
• Instagram dataset file 2
• Instagram dataset file 3