Unraveling Multimodality with Large Language Models.pdf
Skinput
1. Department of Electronics & Communication Engineering
Jyothi EngineeringCollege, Cheruthuruthy
Thrissur – 679531
1
NISHA MENON K
JYAJEEC045
10/4/2013
7. Need for SKINPUT
• Although significant and powerful devices are
being used but their small size typically leads to
limited interaction space
• Skinput supports large interaction space. It
allows users to use their own hands and arms as
touchscreens.
710/4/2013
8. • Skinput is a technology that appropriates the
human body for acoustic transmission.
• It was developed by Chris Harrison, Desney
Tan, and Dan Morris of the Microsoft
Research'sComputational User Experiences
Group(MRCUEG)
• Its first public appearance was at Microsoft's
Tech Fest 2010.
810/4/2013
9. • Giving input through skin.
• Skinput allows the user to simply tap their
skin in order to control audio devices, play
games, make phone calls.
• It uses the sensors to determine where the
user taps on their skin.
910/4/2013
11. 11
Pico-projector
• A very small projector, basically used in
gadgets.
• The system comprises three main parts:
The Laser light source
The Combiner optics
The Scanning mirror
10/4/2013
13. Bio-acoustics
• Study of sound waves inside living body
• When a finger taps the skin, several distinct
forms of energy are produced
• Longitudinal waves
• Transverse waves
• These waves form the integral part of the
whole concept of Skinput
1310/4/2013
ector
14. The signals need to be sensed and worked upon.
This is done by wearing the wave sensor arm
band.
INSIDE VIEWOUTSIDE VIEW
1410/4/2013
15. Bluetooth
• Wireless technology standard
for exchanging data over short distances from
fixed and mobile devices with high levels of
security.
10/4/2013 15
16. The audio
interface digitally
capture data from
sensors
Projector
display image
on arm
vibrations
produced and
passed
through bones
onto skin
Finger tap on arm
then detected
by detector in
armband
1610/4/2013
18. The prototype system is employed with a
MackieOnyx 1200F audio interface to digitally
capture data from the ten sensors.
This is connected via bluetooth to a
conventional desktop computer, where a thin
client written in C interfaced with the device
using the Audio Stream Input/output (ASIO)
protocol
Data is then sent from the thin client over a local
socket to the primary application, written in
Java.
1810/4/2013
19. This program provides a live visualization of
the data from the ten sensors
It segments inputs from the data stream into
independent instances and SVM is used to
classify these input instances
The audio stream is segmented into individual
taps using an absolute exponential average of
all ten channels.
1910/4/2013
20. The software matches sound frequencies to
specific skin locations, allowing the system to
determine which “skin button” the user
pressed.
The prototype system then uses wireless
technology like Bluetooth to transmit the
commands to the device being controlled,
such as a phone, iPod, or computer.
10/4/2013 20
21. FINGERTAPS ON ARM GENERATED SIGNALS
Variation in Bone Density, Size & Mass as well as filtering
effects from SoftTissues & Joints Mean Different Locations
that are acoustically distinct.
2110/4/2013
22. RECOGNIZING INPUT
LOCATIONS
FINALIZING INPUT POINTS
Designed Software Listens for impacts & Classifies them.Then
Different Interactive Capabilities are bounded on different
regions.
2210/4/2013
23. EXPERIMENT
Participants
• 13-> 7 female, 6 male.
• Ages ranged from 20 to 56.
• Body mass indexes (BMIs) ranged
• From 20.5 (normal) to 31.9 (obese).
Each participant was made to memorize the locations for
a minute .
2310/4/2013
24. High BMI is correlated with decreased accuracies.
No direct relation with gender of the participant.
2410/4/2013
25. The projected interface can appear much larger than it ever could
on a device’s screen.
Arm can be brought closer to face (or vice versa) to see the display
close up.
Color contrast can be adjusted by dimming the light so that a
better picture will be visible.
The Skinput could eventually be used without a visual screen. As
the laws of proprioception states, humans are allowed to interact
with specific body parts without using their eyes.This will make
ideal for anyone with little to no eyesight.
The body is portable and always available, and fingers are a
natural input device.
10/4/2013 25
26. One of the current limitations of the prototype is
that the accuracy can degrade over time the longer
you wear it.
A person's Body Mass Index (BMI) will play an
important role in the accuracy of skinput, in obese
people the accuracy rate drops to approximately 80
percent, due to the interference of the wave
transmission by fat deposits in the tissue.
The technology might start up at very high cost
which will not be affordable for the common man.
10/4/2013 26
28. PLAYINGTETRIS: USING FINGERS
AS CONTROL PAD
ACTIVEALSO IN MOVABLE
ENVIRONMENT
Using Fingers,Palms,Arms as Control Any Computing Device
Can be run.Response is real time,robust & remains
functional while walking & Accuracy level is 99.5% 28
10/4/2013
29. The most profound achievement of Skinput is proving that
the human body can be used as a sensor.
A person might walk toward their home, tap their palm to
unlock the door and then tap some virtual buttons on their
arms to turn on theTV and start flipping through channels.
Extensive Research is going on Currently on Skinput to
make the armband more smaller.
Incorporate More Devices withThis System.
Extend accuracy level.
10/4/2013 29
30. With small sized pico-projectors, skinput
oriented systems, are an emerging trend.
Research is carried out for smaller wrist
watch sized sensor arm band
10/4/2013 30
31. How well the Skinput technology works in practice
remains to be seen
The usual factors of performance, price, device
compatibility, and ergonomics still need to be fleshed
out.
The technology itself is intriguing, and may have even
more applications we can't envision yet.
It has been reported this may not appear in commercial
devices for at least 2 years.
Skinput is a very interesting technology. But its fate will
ultimately depend on how committed Microsoft is to
making it a commercial reality and how soon
10/4/2013 31
34. Chris Harrison, DesneyTan, and Dan Morris “Skinput: Appropriating the Skin
as an Interactive Canvas” Microsoft Research 2011.
Chris Harrison, Scott E. Hudson “Scratch Input: Creating, Large Inexpensive,
Unpowered and Mobile Finger Input Surfaces”UIST 2008.
Amento, B.Hill,W.Terveen “The Sound of one Hand: A wrist- mounted bio-
acoustic fingertip gesture interface” CHI’02.
Thomas Hahn “Future Human Computer Interaction with special focus on
input and output techniques” HCI March 2006.
Burges, C.J. ATutorial on SupportVector Machines for Pattern Recognition.
Data Mining and Knowledge Discovery, 2.2, June 1998, 121-167.
10/4/2013 34