The image counterpart to the Ekman's Basic Emotions text scale: the same six cross-culturally recognized categories — anger, disgust, fear, happiness, sadness and surprise — coded from what a face displays rather than from what a speaker says.
Intended for detection studies. Draw a box around each face and assign the emotion category it displays. In a recognition study, use it to mark which affects are present anywhere in the frame. One face may legitimately carry two codes — blends are common, and coding both is more honest than forcing a winner.
What this is not. It is not FACS. FACS codes anatomical action units (AU6 + AU12 and so on) and requires certified training; its manual is licensed and is not reproduced here. This scheme records the emotion category a trained rater perceives, which is a different and coarser judgment — appropriate for reliability work and supervision review, not for claims about specific facial musculature.
Severity is clarity, not intensity. For visual codes the useful question is how legible the display is, not how strongly the person feels: a face at the edge of frame, motion-blurred, or half-occluded is a partial reading no matter how vivid the expression looks. Rate what you can actually see.
Neutral faces are left untagged rather than given a code, so that a neutral reading is distinguishable from a face nobody looked at.