Micro-Behavioral Load Decoding and Spatial Behavioral Signature Recognition in Museums Using Natural Videos

Museum fatigue can weaken visitor experience and learning outcomes, yet post-visit questionnaires and controlled experiments often miss early, low-intensity behavioral changes during natural visits. This study proposes a Micro-behavioral Load Decoding Framework (MLD Framework) linking natural-video collection, micro-behavior recognition, action-group representation, load-direction indicators and spatial behavioral signature recognition. Twelve fatigue-related micro-behaviors were grouped into body-load-related, cognitive-load-related and mixed-load action groups, aggregated as events per 100 visitor observations across eight spatial units, and compared with hall-level questionnaire references. The results show distinguishable combinations of body load, cognitive load and behavioral regulation demand across halls. Some spaces showed low overall load, others showed a dominant load dimension, and others showed simultaneous increases in multiple load-related behaviors. These spatial behavioral load signatures describe relative visitor responses across exhibition environments and reveal spatial couplings among physical exertion, cognitive processing and self-regulation. Natural-video micro-behaviors provide non-invasive process evidence for comparing hall-level load structures and diagnosing exhibition rhythm, information density and restorative support.