<s>
Egocentric	B-General_Concept
vision	I-General_Concept
or	O
first-person	B-General_Concept
vision	I-General_Concept
is	O
a	O
sub-field	O
of	O
computer	B-Application
vision	I-Application
that	O
entails	O
analyzing	O
images	O
and	O
videos	O
captured	O
by	O
a	O
wearable	B-Application
camera	I-Application
,	O
which	O
is	O
typically	O
worn	O
on	O
the	O
head	O
or	O
on	O
the	O
chest	O
and	O
naturally	O
approximates	O
the	O
visual	O
field	O
of	O
the	O
camera	O
wearer	O
.	O
</s>
<s>
The	O
idea	O
of	O
using	O
a	O
wearable	B-Application
camera	I-Application
to	O
gather	O
visual	O
data	O
from	O
a	O
first-person	O
perspective	O
dates	O
back	O
to	O
the	O
70s	O
,	O
when	O
Steve	O
Mann	O
invented	O
"	O
Digital	O
Eye	O
Glass	O
"	O
,	O
a	O
device	O
that	O
,	O
when	O
worn	O
,	O
causes	O
the	O
human	O
eye	O
itself	O
to	O
effectively	O
become	O
both	O
an	O
electronic	O
camera	O
and	O
a	O
television	O
display	O
.	O
</s>
<s>
Subsequently	O
,	O
wearable	B-Application
cameras	I-Application
were	O
used	O
for	O
health-related	O
applications	O
in	O
the	O
context	O
of	O
Humanistic	O
Intelligence	O
and	O
Wearable	O
AI	O
.	O
</s>
<s>
Egocentric	B-General_Concept
vision	I-General_Concept
is	O
best	O
done	O
from	O
the	O
point-of-eye	O
,	O
but	O
may	O
also	O
be	O
done	O
by	O
way	O
of	O
a	O
neck-worn	O
camera	O
when	O
eyeglasses	O
would	O
be	O
in-the-way	O
.	O
</s>
<s>
This	O
neck-worn	O
variant	O
was	O
popularized	O
by	O
way	O
of	O
the	O
Microsoft	B-Device
SenseCam	I-Device
in	O
2006	O
for	O
experimental	O
health	O
research	O
works	O
.	O
</s>
<s>
The	O
interest	O
of	O
the	O
computer	B-Application
vision	I-Application
community	O
into	O
the	O
egocentric	B-General_Concept
paradigm	O
has	O
been	O
arising	O
slowly	O
entering	O
the	O
2010s	O
and	O
it	O
is	O
rapidly	O
growing	O
in	O
recent	O
years	O
,	O
boosted	O
by	O
both	O
the	O
impressive	O
advanced	O
in	O
the	O
field	O
of	O
wearable	B-Operating_System
technology	I-Operating_System
and	O
by	O
the	O
increasing	O
number	O
of	O
potential	O
applications	O
.	O
</s>
<s>
The	O
prototypical	O
first-person	B-General_Concept
vision	I-General_Concept
system	O
described	O
by	O
Kanade	O
and	O
Hebert	O
,	O
in	O
2012	O
is	O
composed	O
by	O
three	O
basic	O
components	O
:	O
a	O
localization	O
component	O
able	O
to	O
estimate	O
the	O
surrounding	O
,	O
a	O
recognition	O
component	O
able	O
to	O
identify	O
object	O
and	O
people	O
,	O
and	O
an	O
activity	B-Application
recognition	I-Application
component	O
,	O
able	O
to	O
provide	O
information	O
about	O
the	O
current	O
activity	O
of	O
the	O
user	O
.	O
</s>
<s>
Following	O
this	O
idea	O
,	O
the	O
first	O
computational	O
techniques	O
for	O
egocentric	B-General_Concept
analysis	O
focused	O
on	O
hand-related	O
activity	B-Application
recognition	I-Application
and	O
social	O
interaction	O
analysis	O
.	O
</s>
<s>
After	O
almost	O
ten	O
years	O
of	O
egocentric	B-General_Concept
vision	I-General_Concept
(	O
2007	O
-	O
2017	O
)	O
,	O
the	O
field	O
is	O
still	O
undergoing	O
diversification	O
.	O
</s>
<s>
Today	O
's	O
wearable	B-Application
cameras	I-Application
are	O
small	O
and	O
lightweight	O
digital	O
recording	O
devices	O
that	O
can	O
acquire	O
images	O
and	O
videos	O
automatically	O
,	O
without	O
the	O
user	O
intervention	O
,	O
with	O
different	O
resolutions	O
and	O
frame	O
rates	O
,	O
and	O
from	O
a	O
first-person	O
point	O
of	O
view	O
.	O
</s>
<s>
Therefore	O
,	O
wearable	B-Application
cameras	I-Application
are	O
naturally	O
primed	O
to	O
gather	O
visual	O
information	O
from	O
our	O
everyday	O
interactions	O
since	O
they	O
offer	O
an	O
intimate	O
perspective	O
of	O
the	O
visual	O
field	O
of	O
the	O
camera	O
wearer	O
.	O
</s>
<s>
The	O
former	O
(	O
e.g.	O
,	O
Narrative	B-Device
Clip	I-Device
and	O
Microsoft	B-Device
SenseCam	I-Device
)	O
,	O
are	O
commonly	O
worn	O
on	O
the	O
chest	O
,	O
and	O
are	O
characterized	O
by	O
a	O
very	O
low	O
frame	O
rate	O
(	O
up	O
to	O
2fpm	O
)	O
that	O
allows	O
to	O
capture	O
images	O
over	O
a	O
long	O
period	O
of	O
time	O
without	O
the	O
need	O
of	O
recharging	O
the	O
battery	O
.	O
</s>
<s>
The	O
latter	O
(	O
e.g.	O
,	O
Google	B-Application
Glass	I-Application
,	O
GoPro	B-Device
)	O
,	O
are	O
commonly	O
mounted	O
on	O
the	O
head	O
,	O
and	O
capture	O
conventional	O
video	O
(	O
around	O
35fps	O
)	O
that	O
allows	O
to	O
capture	O
fine	O
temporal	O
details	O
of	O
interactions	O
.	O
</s>
<s>
Moreover	O
,	O
the	O
camera	O
wearer	O
is	O
not	O
visible	O
in	O
the	O
image	O
and	O
what	O
he/she	O
is	O
doing	O
has	O
to	O
be	O
inferred	O
from	O
the	O
information	O
in	O
the	O
visual	O
field	O
of	O
the	O
camera	O
,	O
implying	O
that	O
important	O
information	O
about	O
the	O
wearer	O
,	O
such	O
for	O
instance	O
as	O
pose	B-Architecture
or	O
facial	O
expression	O
estimation	O
,	O
is	O
not	O
available	O
.	O
</s>
<s>
A	O
collection	O
of	O
studies	O
published	O
in	O
a	O
special	O
theme	O
issue	O
of	O
the	O
American	O
Journal	O
of	O
Preventive	O
Medicine	O
has	O
demonstrated	O
the	O
potential	O
of	O
lifelogs	O
captured	O
through	O
wearable	B-Application
cameras	I-Application
from	O
a	O
number	O
of	O
viewpoints	O
.	O
</s>
<s>
More	O
recently	O
,	O
egocentric	B-General_Concept
cameras	O
have	O
been	O
used	O
to	O
study	O
human	O
and	O
animal	O
cognition	O
,	O
human-human	O
social	O
interaction	O
,	O
human-robot	O
interaction	O
,	O
human	O
expertise	O
in	O
complex	O
tasks	O
.	O
</s>
<s>
Other	O
applications	O
include	O
navigation/assistive	O
technologies	O
for	O
the	O
blind	O
,	O
monitoring	O
and	O
assistance	O
of	O
industrial	O
workflows	O
,	O
and	O
augmented	B-General_Concept
reality	I-General_Concept
interfaces	O
.	O
</s>
