Task 1. Grounded Affect Recognition
Given the focal subject and episode, infer the subject's affective state and its manifestation from the video.
Fields
subject: the focal person in the affective episode. Its value is given.
affective_state: what the subject feels.
object: the person toward whom the affect is directed, identified by label or by description.coarse_category: the category of the subject's affective state, as a list of one or more of the 12 categories Anger, Anxiety, Disgust, Embarrassment, Fear, Frustration, Guilt, Happiness, Relief, Sadness, Surprise, Neutral.fine_description: free-form text giving the contextual nuance or blend of the affect.polarity: the valence of the affective state, one of Positive, Negative, Neutral, Mixed. Mixed means that positive and negative feelings coexist in the same person at the same moment. A mismatch between the affective state and its manifestation, or a change across moments, is not sufficient.intensity: the strength of the affective state, one of High, Medium, Low, None. Low is a mild feeling with little influence on attention or ongoing activity. Medium is a feeling that changes attention or response without dominating the episode. High is a strong feeling that dominates the person's concerns or creates an urgent need to respond.
affective_manifestation: the outward expressions corresponding to the affective state, each in free-form text.
facial_expression: observable facial movement or configuration.body_language: observable posture, gesture, movement, or spatial behavior.vocal_cues: the acoustic and prosodic delivery of the subject's speech, such as tone, pitch, volume, pace, or rhythm.verbal_content: the subject's words or speech act.
Output
The answer is one JSON object with the fields of the answer template in the question. Placeholder entries in the template denote values to infer. The subject, object and recipient identifiers supplied in the template are fixed.