05.03.2019 Views

March 2019

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

20 <strong>March</strong> <strong>2019</strong><br />

TECH<br />

www.theasianindependent.co.uk<br />

Facial recognition software to help<br />

identify US Civil War soldiers<br />

WASHINGTON : Scientists<br />

have developed a software that<br />

uses crowdsourcing to help<br />

algorithms identify faces in<br />

photos, that could uncover the<br />

mysteries of the nearly 4 million<br />

photographs of Civil Warera<br />

images that may exist in the<br />

historical record. Kurt Luther,<br />

an assistant professor at<br />

Virginia Tech in the US, was<br />

inspired to develop the software<br />

for Civil War Photo<br />

Sleuth while visiting the Heinz<br />

History Center's exhibit called<br />

"Pennsylvania's Civil War" in<br />

Pittsburgh, Pennsylvania.<br />

There he stumbled upon a Civil<br />

War-era portrait of Oliver<br />

Croxton, his great-great-great<br />

uncle who served in Company<br />

E of the 134th Pennsylvania,<br />

clad in a corporal's uniform.<br />

"Historical photos can tell us<br />

a lot about not only our own<br />

familial history but also inform<br />

the historical record of the time<br />

more broadly than just reading<br />

about the event in a history<br />

book," said Luther.<br />

The Civil War Photo Sleuth<br />

project allows users to upload<br />

photos, tag them with visual<br />

cues, and connect them to profiles<br />

of Civil War soldiers with<br />

detailed records of military history.<br />

Photo Sleuth's initial reference<br />

database contained more<br />

than 15,000 identified Civil<br />

War soldier portraits from public<br />

domain sources like the US<br />

Military History Institute and<br />

other private collections.<br />

More than 600 users contributed<br />

more than 2,000 Civil<br />

War photos to the website in<br />

the first month after the launch,<br />

and roughly half of those photos<br />

were unidentified.<br />

Over 100 of these unknown<br />

photos were linked to specific<br />

soldiers, and an expert analysis<br />

found that over 85 percent of<br />

these proposed identifications<br />

were probably or definitely<br />

NASA decodes how our Moon got 'sunburns'<br />

WASHINGTON : The<br />

Moons 'sunburns'—distinctive<br />

patterns of swirls—are a result<br />

of interactions between the<br />

Sun's damaging radiation with<br />

pockets of lunar magnetic field,<br />

according to NASA data. Every<br />

object, planet or person travelling<br />

through space has to contend<br />

with the Sun's damaging<br />

radiation. Research using data<br />

from NASA's Acceleration,<br />

Reconnection, Turbulence and<br />

Electrodynamics of the Moon's<br />

Interaction with the Sun<br />

(ARTEMIS) mission suggests<br />

that lunar swirls could be the<br />

result of solar wind interactions<br />

with the Moon's isolated pockets<br />

of magnetic field.<br />

The Sun releases a continuous<br />

outflow of particles and<br />

radiation called the solar wind.<br />

The solar wind washes over the<br />

planets, moons and other bodies<br />

in our solar system, filling a<br />

bubble of space—called the<br />

heliosphere—that extends far<br />

past the orbit of Pluto. On<br />

Earth, we are largely protected<br />

from the damaging effects of<br />

the solar wind. Since the solar<br />

wind is magnetised, Earth's natural<br />

magnetic field deflects the<br />

solar wind particles around our<br />

planet so that only a small fraction<br />

of them reach our planet's<br />

atmosphere.<br />

Unlike Earth, the Moon has<br />

no global magnetic field.<br />

However, magnetised rocks<br />

near the lunar surface do create<br />

small, localised spots of magnetic<br />

field that extend anywhere<br />

from hundreds of yards<br />

to hundreds of miles.<br />

This is the kind of information<br />

that needs to be well<br />

understood to better protect<br />

astronauts on the Moon from<br />

the effects of radiation,<br />

researchers said. The magnetic<br />

field bubbles by themselves are<br />

not robust enough to protect<br />

humans from that harsh radiation<br />

environment, but studying<br />

their structure could help<br />

develop techniques to protect<br />

our future explorers.<br />

"The magnetic fields in<br />

some regions are locally acting<br />

as this magnetic sunscreen,"<br />

said Andrew Poppe, a scientist<br />

at the University of California,<br />

Berkeley in the US.<br />

These small bubbles of magnetic<br />

"sunscreen" can also<br />

deflect solar wind particles—<br />

but on a much smaller scale<br />

than Earth's magnetic field.<br />

While they are not enough to<br />

protect astronauts by themselves,<br />

they do have a fundamental<br />

effect on the Moon's<br />

appearance. Under these miniature<br />

magnetic umbrellas, the<br />

material that makes up the<br />

Moon's surface, called regolith,<br />

is shielded from the Sun's particles.<br />

As those particles flow<br />

toward the Moon, they are<br />

deflected to the areas just<br />

around the magnetic bubbles,<br />

where chemical reactions with<br />

the regolith darken the surface.<br />

This creates the distinctive<br />

swirls of darker and lighter<br />

material that are so prominent<br />

they can be seen from Earth—<br />

one more piece of the puzzle to<br />

help us understand the neighbour<br />

NASA plans to re-visit<br />

within the next decade.<br />

correct. Presently, the database<br />

has grown to over 4,000 registered<br />

users and more than 8,000<br />

photos. "Typically, crowdsourced<br />

research such as this is<br />

challenging for novices if users<br />

don't have specific knowledge<br />

of the subject area," said<br />

Luther. "The step-by-step<br />

process of tagging visual clues<br />

and applying search filters<br />

linked to military service<br />

records makes this detective<br />

work more accessible, even for<br />

those that may not have a deeper<br />

knowledge of Civil War military<br />

history," he said.<br />

Person identification tasks<br />

can be challenging in larger<br />

candidate pools because there<br />

is a larger risk for false positives.<br />

The novel approach behind<br />

Civil War Photo Sleuth is based<br />

on the analogy of finding a needle<br />

in a haystack. The data<br />

pipeline has three haystackrelated<br />

components: building<br />

the haystack, narrowing down<br />

the haystack, and finding the<br />

needle in the haystack.<br />

When combined, they allow<br />

users to identify unknown soldiers<br />

while reducing the risk of<br />

false positives. Any time a user<br />

uploads a photo to identify it,<br />

the photo gets added to the<br />

site's digital archive or<br />

"haystack," making it available<br />

for future searches.<br />

Following upload, the user<br />

tags metadata related to the<br />

photograph such as photo format<br />

or inscriptions, as well as<br />

visual clues, such as coat<br />

colour, chevrons, shoulder<br />

straps, collar insignia, and hat<br />

insignia. These tags are linked<br />

New Delhi : WhatsApp is<br />

reportedly working on a new<br />

‘Advanced Search’ feature that<br />

will be available for iOS beta<br />

testers soon. According to a<br />

WABetaInfo report, the feature<br />

will let users search for all<br />

types of messages including<br />

photos, links, audio, GIFs, and<br />

more in chat.<br />

WhatsApp currently allows<br />

for text search through its<br />

search feature. WhatsApp currently<br />

allows for text search<br />

through its search feature. It<br />

looks like the ‘Advanced<br />

Search’ feature will let users<br />

search for more like photos,<br />

GIFs, links, documents, videos,<br />

documents, and audio.<br />

The Advanced Search feature<br />

will also display the number<br />

of media files stored on the<br />

user’s device segregated as<br />

photos, videos, GIFs, audio,<br />

etc. When one clicks on a specific<br />

type of media, all the messages<br />

containing that media are<br />

displayed.<br />

It will also show a preview<br />

of the media. So for instance,<br />

when users type photos in<br />

WhatsApp’s Advanced Search<br />

bar, they will be shown a preview<br />

of each image in a listicle<br />

format. The search can be further<br />

filtered by typing a specific<br />

word for the exact photo<br />

users are looking for.<br />

As per the report,<br />

to search filters to prioritise the<br />

most likely matches. For example,<br />

a soldier tagged with the<br />

"hunting horn" hat insignia<br />

would suggest potential matches<br />

who served in the infantry,<br />

while hiding results from the<br />

cavalry or artillery.<br />

Next, the site uses state-ofthe-art<br />

face recognition technology<br />

to eliminate very different-looking<br />

faces and sort the<br />

remaining ones by similarity.<br />

Both the tagging and face<br />

recognition steps narrow down<br />

the haystack.<br />

Finally, users find the needle<br />

in the haystack by exploring the<br />

highest-probability matches in<br />

more detail. A comparison tool<br />

with pan and zoom controls<br />

helps users carefully inspect a<br />

possible match and, if they<br />

decide it's a match, link the previously<br />

unknown photo to its<br />

new identity and biographical<br />

details. Retracing historical<br />

Civil War photos through facial<br />

recognition software like Photo<br />

Sleuth has broad applications<br />

beyond identifying historical<br />

photos, too. The software has<br />

the potential to generate new<br />

ways to think about building<br />

person identification systems<br />

that look beyond face recognition<br />

and leverage the complementary<br />

strengths of both<br />

human and artificial intelligence.<br />

WhatsApp iOS beta testers to get<br />

‘Advanced Search’ feature soon<br />

WhatsApp is reportedly working on a new 'Advanced Search'<br />

feature that will be available for iOS beta testers soon. The<br />

feature will let users search for all types of messages including<br />

photos, links, audio, GIFs, and more in chat.<br />

WhatsApp’s ‘Advanced<br />

Search’ feature is under development<br />

at this moment and is<br />

not available. It will be rolled<br />

out for iOS beta testers initially<br />

and Android availability is<br />

expected to follow. In the<br />

development phase, the audio<br />

preview has not been enabled,<br />

neither it is known whether it<br />

will be possible to search for<br />

Locations and vCards using<br />

this feature.

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!