Research steps towards human sequence evaluation

Abstract

Human Sequence Evaluation (HSE) concentrates on how to extract descriptions of human behaviour from videos in a restricted discourse domain, such as (i) pedestrians crossing inner-city roads where pedestrians appear approaching or waiting at stops of busses or trams, and (ii) humans in indoor worlds like an airport hall, a train station, or a lobby. These discourse domains allow exploring a coherent evaluation of human movements and facial expressions across a wide variation of scale. This general approach lends itself to various cognitive surveillance scenarios at varying degrees of resolution: from wide-field-of-view multiple-agent scenes, through to more specific inferences of emotional state that could be elicited from high resolution imagery of faces. The true challenge of HSE will consist of the development of a system facility which starts with basic knowledge about pedestrian behaviour in the chosen discourse domain, but could cluster evaluation results into semantically meaningful subsets of behaviours. The envisaged system will comprise an internal logic-based representation which enables it to comment each individual subset, giving natural language explanations of why the system has created the subset in question.