Player is loading...
LaternaMagica
Wednesday, Aug. 24, 2022 · 1:08 p.m. · 10m 56s
LaternaMagica is an interactive beamer that can project on different surfaces in an indoor environment. The interaction capability will be achieved through natural gestures and gaze by using different existing AI techniques. The proposed device will be conceived with a motorized mirror coupled with a beamer/projector and an RGBD camera.
Embed
Transcriptions
Note: this content has been automatically generated.
00:00:00
yeah i like that analogy car i'm seeing michael you know i i
00:00:07
let's travel back in time seventeen seventy six this is one of the first demonstrate off a projectile
00:00:15
and if we look at what we have today things have not changed so much in the sense that technology's better
00:00:22
but the main component of the project all remains the same something
00:00:27
very static but we play somewhere on but really the only interact with
00:00:34
so we would like to see the project toss on more as more dynamic on more interactive
00:00:40
with latin america we're transforming salt project tossed into interactive orbits
00:00:48
this whole bought can understand its environment and the surrounding
00:00:52
so we need first to be now we can just come down you know as a platform
00:00:56
then we have a knack treated me all in order to explore
00:01:00
it to the environment on the the wall space will be displaced
00:01:04
and then we have a cover or what is used both to understand the surrounding but also
00:01:09
to pipe we used uh in order to interact with the user on provide new communication capability
00:01:16
overworked cost of what of this for the new components that we ah
00:01:21
he is very low compared to the price of a bimbo meaning but
00:01:25
it's not so difficult to improve the be locked up i. b. t.'s nobody's
00:01:31
so what do we need to do that to behind this we need to understand do you in
00:01:35
which the you know he's based on for this while using the single cover all but his first
00:01:42
brought by the user on the user the scanning the one put back the cover on the device
00:01:47
and we just got the device can understand well towards well
00:01:51
the planets of faces but you can use for the projections
00:01:56
and then the system is automatically segmenting planes finding well these productions of feces
00:02:02
on backtracking this got it on we can vent yeah what do you use a or
00:02:06
pointing at this difference of feces and deciding which of phase should be used for the projection
00:02:18
then we have this activation capability so it means that if i'm showing to
00:02:23
the system well actually project different sorts user can be different part of the presentation
00:02:29
then we have the device automatically moving from one production to yourself just
00:02:33
can be table artist can you watch this can be the war surrounding
00:02:39
and you risk your life the more of this just stuff though so this is the backup plan how are you
00:02:48
so we believe that this technology can be used with menus uh whoops
00:02:53
so first that analogy cut can be used at the shop floor in order to project information on objects
00:03:00
to add to the walk else especially if the all these devices placed
00:03:04
on the mobile platform by projecting on initial faces on updating does dynamically
00:03:11
then since we know where the use of these we can also a few yeah fact being
00:03:14
projected in the sense that we can then the l. three d. object but we knew each
00:03:19
as a very uh three d. why they are flat but if we know where
00:03:22
the user is we get at this kind of animal fake effect that we can explore
00:03:27
so it's uh in in a series of experience by using this on with that reality capability
00:03:33
on without worrying special glasses of special and nets all without having to and er lu device because
00:03:39
it's projected so it means that the user can continue the activity without taking too under another device
00:03:48
so we believe it can be using public space used for easy mission fails all schools events all musicians
00:03:55
but it can be used at all if we want to keep places multiple screens
00:04:00
on instead of a single project all i can project indifference bases
00:04:06
all that watch the smart meeting hoops and we splendidly presentations on this is the use case
00:04:12
but we re for all that we have followed during devices so we would like to avoid that
00:04:18
and it's yeah we can do this so there is a for the presentation content but there is also the menu and it's like
00:04:23
this communication medium and let's see what what we can do with this
00:04:26
communication medium today to do that we have this to interact with the device
00:04:32
if you think of what it is it's a point to that point but i can we go on the screen
00:04:38
and it's two buttons will fall while to go back well and if i need to go tense then slides but well that we pressed ten
00:04:44
times this one point three billion can do better by using the space to have something more tangible or to to exploit the space where we'll
00:04:54
so we require support fees to do the scene understanding use
00:04:57
of tracking gesture recognition about code for make at twenty to design
00:05:02
we need the full production mapping in order to distort the way the images of project that
00:05:08
and we to media design to have new contents for these new type of devices
00:05:13
and there's a hoop real we have all of these uh expect
00:05:17
dies as a as a team reason very distinct uh expert flies
00:05:23
so i i think the hoboken interaction hoop with uh no all we know buckle for optimisation
00:05:30
gain id is uh recently uh uh some of the stuff that
00:05:34
could scan terrific on user so years or so ago to channel
00:05:37
with the same name and use them or cut ties in the use of habits which make up a new designs on feeding in printing
00:05:44
and then you are from birds miss universe joe
00:05:46
front on with expectation digital ah visualisation on multimedia
00:05:51
so we believe well the right team to be this
00:05:54
kind of multi uh of milk incompetence uh the device
00:05:59
now if we look at the market it looks interesting marquette because if
00:06:02
we take a single technology for emails that there's a disco d. a. g.
00:06:06
from texas instruments it's already representing five billions units being so or
00:06:12
uh immediate twenty twenty and we isn't usually five percent expected drove
00:06:20
and you know if you look at the competitors so first to
00:06:23
their production markings that are quite popular our production marketing is only fixed
00:06:30
it's a tiny realisation so it's something really that is
00:06:34
don't completely off line it's more like making a movie
00:06:37
so the the business model is very different compared to what we want to do on is not interactive
00:06:44
then there is this whole school i chew system but is mostly done for lights
00:06:49
not projecting a lot like that content but just projecting some light on
00:06:53
the seat and also not interacting the closest would be mule as though
00:06:59
middle head is also using a new hole you can also use of our project alls
00:07:04
but it's only peeping object all the meaning but there is no interaction
00:07:09
and it's also business model where all the everything is known
00:07:13
by the company and then they ran to the device while we
00:07:15
would like the users to create a on content we would like uses to use this kind of device the way they want
00:07:24
we have several things to do next refinished a prototype during the night just that there
00:07:29
so we still have things that we would like to finish there is the continues interaction aspect
00:07:35
there is the gaze talking functionality that we would like to ah we have a new prototype on going with a
00:07:41
moving cover on using exactly the same mechanism but don't upside down on the the colour also being able to move
00:07:49
finally we need to think about the use of software the i thought open train users tune to use these new devices
00:07:55
since we have this new uh movement capability we also
00:08:00
need to find new ways of training the person's to use
00:08:04
and if we go forward with this uh views he will have to find what is the market we want to target
00:08:09
first as there is no these multiple use uh whoops and
00:08:13
ask end users what they would like this device to do
00:08:18
so we're latin america we have transforming but project also into interactive robots
00:08:23
well now we will try to see of prototype inaction on your all looking for for for this uh
00:08:40
yeah
00:08:51
i
00:09:01
uh_huh
00:09:12
uh_huh
00:09:20
in the worst case with a wire frames about being this
00:09:25
ah ah ah ah
00:09:33
oh okay
00:09:46
while trying uh we we we use the the back the backup or
00:09:57
oh
00:10:00
something with manually flashing on g. huh
00:10:06
a a a a a yeah
00:10:15
yeah
00:10:23
yeah
00:10:40
right off of accent things that things than that they're there for a more rapid wanted
00:10:51
okay
Conference Program
LightAI4Comfort
Aug. 24, 2022 · 12:13 p.m.
119 views
LaternaMagica
Aug. 24, 2022 · 1:08 p.m.
118 views
SolveHR: Natural language inference for professional search
Aug. 24, 2022 · 1:26 p.m.
Recommended talks
LeukoCounter Pitch
Gabriel Minoru
Sept. 17, 2018 · 4:53 p.m.
135 views
