10. Competition and positioning OMG, Google is doing the same thing
No competition, no market
Lear
ning
s fr
om fo
undi
ng a
Com
pute
r Vis
ion
Star
tup
Basic competition check (again)
www.crunchbase.com
Lear
ning
s fr
om fo
undi
ng a
Com
pute
r Vis
ion
Star
tup Differentiation
How are you different from your competition?(performance, features, geographic, price, business model, ...)
Lear
ning
s fr
om fo
undi
ng a
Com
pute
r Vis
ion
Star
tup
How to handle competition?
Be aware of who competitors are and what they do (you need to be able to answer when people ask)
But...Focus on your own ideas and on making your product great
Be inspired if competition does something good but don’t copy or follow blindly - copying lacks understanding, better think for yourselfCompetitors might have a different agenda and goals
And...Don’t worry - a reasonable amount of competition is good for all
Lear
ning
s fr
om fo
undi
ng a
Com
pute
r Vis
ion
Star
tup
OMG, what happens if Google is doing “the same”?
- they never really do the exact same thing- they most likely have a different goal- they most certainly have a different business model- they have competitors who might want partners (you)
Lear
ning
s fr
om fo
undi
ng a
Com
pute
r Vis
ion
Star
tup
Polar Rose: How we did it
Case 1: Picasa & iPhotoGoogle Picasa added face recognition for people tagging in fall 2008
Apple’s iPhoto added the same in January 2009
At the same time we were doing people tagging at polarrose.com (Flickr/Facebook)
End result: Picasa and iPhoto drive competitors to feature parity
Lear
ning
s fr
om fo
undi
ng a
Com
pute
r Vis
ion
Star
tup
Polar Rose: How we did it
Case 2: GogglesGoogle Goggles launched late fall 2009 with supposed face recognition feature disabled
At the same time we had the Augmented ID / Recognizr app
End result: still unclear, but large players are more careful, small startups can take risks and provoke. Again a drive for feature parity elsewhere is emerging.
Lear
ning
s fr
om fo
undi
ng a
Com
pute
r Vis
ion
Star
tup
A brief history of visual search and kooaba
Q1/08 Q2/08
Amaz
on a
cqui
res
snap
tell
Q3/08 Q4/08
“Tra
ctio
n”
Wor
lds
first
iPho
ne a
pp
for v
isua
l sea
rch
Q1/09 Q2/09 Q3/09 Q4/09 Q1/10Q4/07Q3/07Q2/07
kooa
ba T
ech
Talk
@
Goo
gle(
USA
)
first
iPho
ne
iPho
ne S
DK
MM
S ca
mpa
ign
with
eas
yjet
first
ver
sion
vis
ual s
earc
h
> 1
milli
on it
ems
in D
B
> 10
milli
on it
ems
in D
B
$$ Seed $$ Angel
Lear
ning
s fr
om fo
undi
ng a
Com
pute
r Vis
ion
Star
tup
References
Books:ReWork (again)
Round up
Lear
ning
s fr
om fo
undi
ng a
Com
pute
r Vis
ion
Star
tup
Things we didn’t cover
Some boring parts:IP & patents
How to handle admin stuff (bank, bookkeeping, legal,...)
Office & finding a good place to workB2B licensing alternatives and structures How to do consultancy work
Deeper into certain topics we presented today.
We could tell you next CVPR if you want ;)
Lear
ning
s fr
om fo
undi
ng a
Com
pute
r Vis
ion
Star
tup
Take home messages
Starting out is cheaper and easier than you think
Focus on core functionality and engage customers early
Vision is about to enter consumer market, timing is good
Lots of the things we learned and told you today turn out to be “just” (advanced) common sense.
Lear
ning
s fr
om fo
undi
ng a
Com
pute
r Vis
ion
Star
tup
Vision specific lessons
Vision is magic to many, explanations often needed
Complex technology (you will need funding / quality in product development)
Business models in B2C often unclear (innovation is needed here, too)
But huge opportunities for you as an expert, if you’re not a total techie ;)
Feedback?
The End
Till [email protected]
@tmdqwww.quack.ch
Jan Erik [email protected]
@jesolemwww.janeriksolem.net
Slides will be online!http://www.vision.ee.ethz.ch/tquack/cvpr10-tutorial.html