Segmenting Supervised Activities In A Video Sequence Based On Handling Of Artifacts Towards Intelligent Systems

Martínez-Pérez, F.E., Pérez-González, H.G., González-Fraga, J.A., Cuevas-Tello, J.C. and Nava-Munoz, S.E.
Research in Computing Science IPN, 68(1), pp. 108-119, 2014.


Nowadays intelligent systems community has conducted research on human activity recognition. For example, in a healthcare environment, we would like to know what activities (feeding, blood pressure, hygiene and medication) are performed by a caregiver given a video sequence (recorded by a surveillance system). Specifically, it is complicated to infer those activities that are performed using one or several artifacts at different times, so the activity inference and video segmentation are complex tasks. Additionally, it is desirable to perform the video segmentation in an automatic fashion. Therefore, in this paper we present an intelligent system for video segmentation. We present an example in a realistic environment in which the analysis of video sequences per day was reduced by using video segmentation.