2. Technology
• Image
recogni1on
(IR)
–
technology
for
automa1c
iden1fying
object
on
the
image
• The
technology
can
be
widely
used
in
retail
to
iden1fy
consumer’s
demands
Our
product
• Image
recogni1on
technology
implementa1on
to
mobile
devices.
• Mobile
apparel
e-‐commerce
2
3. The
problem
• Enormous
variety
of
clothing
e-‐retailers
• Clothing
search
becomes
1me-‐consuming
• Minority
of
e-‐tailing
web
sites
compa1ble
with
mobile
devices
– Less
than
2%
conversion
rate
from
mobile
devices
Solu1on
• Easy
to
search
• Consumer
needs
only
a
photo
of
the
clothes
taken
from
– Magazine
– Internet
– Private
photo
album
• The
result
relates
to
the
query
by:
– Shape
– Texture
– Colour
3
4. How
it
works
Ini1al
photo
Segmenta1on
• Detec1ng
the
object
Searching
• Searching
in
the
Database
Result
• User
can
improve
the
result
by
colour
or
shape
Purchase/
Follow/
Wishlist
• «Follow»
–
user
can
follow
the
query
if
a
desired
object
is
not
found
4
5. Search
results
(demo
example)
Match
by
colour
and
shape
Query
Match
by
shape
Match
by
colour
or
shape
Match
by
shape
Arrow2
Arrow2
5
“Arrow
2”
–
result
improvement.
The
resul1ng
object
can
become
next
query
object
6. The
business
6
Ini1al
image
purchase
revenue
Happy
customer
Mobile
app
7. Applica1on
of
similar
technology
Company
Name
Image
Recogni0on
Image
Recogni0on
on
Mobile
Pla5orm
Shape,
colour,
texture
analysis
(clothes)
Closest
offer
to
the
query
Sale
from
the
pla5orm
Google
glass*
✓
✗
✓
✗
✗
eBay**
✓
✓
✗
✗
✓
Amazon***
✓
✓
✗
✗
✓
Hunga
✓
✓
✓
✓
✓
*
Announced
**
“eBay
motors
app”
***Amazon’s
image
recogni1on
only
on
books
7
8. Mobile
device
traffic
• Amazon.com
-‐
27%
traffic
from
mobile
devices
in
February
2013
-‐
comScore,
• eBay
-‐
29%
traffic
from
mobile
in
February
2013.
• 25%
-‐
Apparel
purchase
on
eBay
8
9. US
US
M-‐commerce
apparel
sales
via
smartphones,
2012-‐2017
year
2012
2013
2014
2015
2016
2017
sales,
bln
$
8
12
17
22
27
31
m-‐commerce
share
3%
5%
6%
7%
8%
9%
RUS
RUS
M-‐commerce
apparel
sales
via
smartphones,
2012-‐2017
year
2012
2013
2014
2015
2016
2017
sales,
bln
$
0,36
0,84
1,08
1,6
1,9
2,58
m-‐commerce
share
3%
4%
4%
5%
5%
6%
9