SEMAINE Emotion Recognition Video Dataset

This dataset is a video dataset that captures the audio-visual interaction between users and operators. The operators play characters with four personalities, namely Poppy (happy), Obadiah (gloomy), Spike (angry), and Prudence (prudent). The video is recorded at a rate of 25 fps (352 x 288 pixels), and the video clips are annotated into several states, such as agreement, interest, certainty, concentration, and deliberation, and are continuously scored in the range of [-1,1], where -1 represents the lowest evaluation, that is, completely inattentive, and +1 represents the highest evaluation, that is, the most focused.

In the Solid SAL scenario, 24 recording sessions are used. The recordings consist of two parts: user and operator. Each recording usually has 4 character interactions, providing a total of 95 character interactions and 190 video clips.