[evaal contest] Idea about improvements for the next year competition

[Dario Salvi]

Dear competitors,


we have had an internal discussion about how to improve next year's 
competition. The principle is that we would like to push the competition 
towards comprehensive context awareness, but we don't want to loose the 
participation of those who work on the technologies behind it, like 
localization.

As a first stepin that direction, we would like to merge activity 
recognition and localization, allowing those who are keen only on 
localization to compete anyway.

To get the idea, the competitors will have to provide us with:

- precise X,Y location of the actor

- similarly to the AOIs the areas: "bedroom", "kitchen", "bathroom", and 
also more precise locations like: "on bed", "in front of fridge", "at 
the door"

- a simple movement classification like: "still", "laying", "walking", 
"running", "cycling", and  "falling" (yes, it would be nice to have fall 
detection)

- in one of the benchmarks we would also like to locate two actors 
instead of one


The idea is that those system that can recognize some movements (e.g. 
with accelerometers) would get a higher scoring, but localization alone 
would still be the main thing to be evaluated.

What I would like to ask you is if you find the challenge interesting, 
or if you think it's too hard or out of the scope.

Any feedback is warmly welcome.




Best regards,


Dario

-- 
Dario Salvi
Researcher in AAL and personalized health systems

https://orcid.org/0000-0002-9203-1124
http://www.researcherid.com/rid/E-5754-2012
http://scholar.google.es/citations?user=Yeoj3BEAAAAJ

Current affiliation:
Life Supporting Technologies
Departamento de Tecnología Fotónica y Bioingeniería
Universidad Politécnica de Madrid
Madrid, Spain