COVER SMARTPHONES
Left: One of the many
problems AI can solve
is to make intelligent
conclusions from
images or videos
www.newelectronics.co.uk 9 March 2021 11
9parusnikov/stock.adobe.com
“AI can be used
to enhance
traditional
applications.
This is especially
true with the
explosion in AIprocessing
power
in new chipsets”
Johan Svensson
by Apple which launched the A14 chip
and M1 Macbook chip at the end of
last year.
Artificial Intelligence
Articial intelligence is now present
across a growing number of
smartphones. While it had traditionally
been reserved for the most expensive
devices it can now be found in more
affordable handsets, delivering
important new functions such as
voice assistants, photography, object
recognition and faster data processing.
AI can be deployed to help users by
learning their habits and tastes and
then delivering suggested actions
through different applications.
AI is able to mimic the intelligence
or behavioural patterns of humans and
can be used to solve problems that
traditional computing has had difculty
in doing in the past.
The Huawei Mate 10 was the rst
mobile device to feature AI and today’s
smartphones using AI are better able
to manage device performance and
respond more accurately to their
users’ needs.
The impact of Al on the smartphone
market has already been signicant
with Amazon’s Alexa, or Apple’s Siri AIbased
voice assistants. By interacting
with devices in other ways than
through a graphical interface it is now
possible for new applications to take
shape.
“One of the many problems AI
can solve is to make intelligent
conclusions from images or videos
such as facial recognition, object
detection and tracking, motion
detection and motion vector analysis,”
explained Johan Svensson, CTO
of Imint, a Swedish company that
specialises in intelligent sensor and
data analysis.
“At Imint, we believe that with
these building blocks there are even
more possibilities for new applications
within smartphones. We have
already seen several applications
such as image segmentation, face
beautication, in addition to VR and
AR applications like furnishing your
home with virtual furniture, or trying on
new clothes with virtual clothing.
“AI can be used to enhance
traditional applications. This is
especially true with the explosion in
AI-processing power in new chipsets.
That said, solving problems usually
requires enormous amounts of data so
that algorithms can learn to perform
well, which puts certain companies,
such as Google, in a strong position.”
According to Svensson, AI can be
used to address many different areas
and applications.
“The specic application that
pertains to Imint – and the best way to
describe it – is that it’s an extremely
powerful technology that enables
improved real-time video enhancement
processing on a device in a way that
anticipates the user’s videography
desires and/ or intent – and then autoapplies
the appropriate adjustments to
produce improved content.
“We’ve already been integrating AI
into Vidhance - our agship product
suite. The Vidhance Sele Mode
combines AI and facial recognition to
help ensure a user stays in-frame when
video chatting or recording Instagram
stories, for example.”
By integrating AI and ML directly
into smartphone software Svensson
said it was possible to create more
compelling video content and enjoy
richer communication with friends,
family, and co-workers.
“The ideal scenario for an enduser,
when it comes to making a
video, is when the technology knows
what the user wants beforehand, and
makes the requisite adjustments and
corrections,” Svensson suggested.
This could include technical
adjustments, like auto-correcting a
horizon or the auto-removal of noise
and discolorations, or delivering
creative adjustments, like auto-framing
video subjects or auto-applying lters
and effects.
“AI can also be used in up-scaling
or super-resolution where users want
to have many more details than what’s
actually physically captured by the
sensor.”
But that’s only one of many benets
from using AI.
“There are a variety of benets to
the end-user. For one, the user will be
able to have access to many new types
of applications and features that will
be able to gather even more types of
information and intelligence.”
The roll-out and availability of 5G
will accelerate the use of AI, according
to Svensson.
“With the bandwidth and low
latency that 5G promises, we can start
redening the way in which network
infrastructure is developed. Previously,
with less bandwidth, many applications
have been required to do calculations
/www.newelectronics.co.uk
/stock.adobe.com