Describing events

Activity Recognition using an ontology-base language

Participants: Carlos Crispim, Francois Bremond.

The page presents a general video understanding approach for the description and recognition of events occurring in videos, for real world (home-care) applications.

Here below are two videos showing 1) a raw video recorded at Thessaloniki pilot (the 4-activity protocol) and 2) the video processed with INRIA activity recognition system using as input a file describing the events to be recognized thanks to an ontology-base language defined in the team. The goal of this demonstration is for you to propose your own description of the events using the same langage.

Figure 1. Test Video from pilot in Thessaloniki dipicting an indivudal, who is performing various tasks of everyday life.

To describe which activity you want to recognize: Here is a file you can use to describe the activities

To start the Demo: Here is the link for the demo

Below you may find also demos about 3D activity recognition using color-depth sensors. If you want to try our system for non-commercial purposes, feel free to check this link Complex Activity Recognition system

3D activity recognition in GAADRD data set (public data set, Request access)

3D activity recognition in ICP institute (Our approach is used in a daily basis by doctors)

3D activity recognition in Nursing home: 24/7 monitoring of frailty patterns.