In the field of video-based emotion recognition, the analysis of micro and macro behaviors is essential. Micro behaviors, such as eyelid movements, subtle facial expressions, and fine muscular action units, reveal hidden affective states. For example, anger may be indicated by tightened lips or narrowed eyes, while sadness might manifest in lowered corners of the mouth.

Macro behaviors provide complementary insights, including posture, hand gestures, and locomotor patterns such as pacing or withdrawal. Beyond these, contextual cues – like the presence of potentially dangerous objects (e.g., knives) – offer critical information for assessing situational risks and behavioral intent.

Advancements in automation now allow computer systems to integrate these multimodal signals into robust recognition models. Research demonstrates that micro-level facial muscle activity not only reflects different emotions but can also serve as markers for psychological disorders. At the same time, macro-level behaviors and object detection enrich the interpretative context, enabling more comprehensive and reliable affective computing solutions.