Martin Calmels, Geopost Vision: street-level imagery from delivery vans

1
Hello from the Kurrant booth.

2
We are here with Martin Calmels.

3
He is the CEO of Geopost Vision,
and we're going to be talking

4
about moving data, but moving data
in the sense of sensors moving.

5
Can you tell us a bit about your
solution to start the conversation?

6
Yeah, thank you very much.

7
So first of all,
Geopost Vision is part

8
of a very large group called Geopost.

9
Geopost is the largest parcel
delivery network in Europe,

10
meaning that our job is the last
mile delivery, delivering parcels

11
to everyone every day in more
than 20 countries in Europe.

12
So what we are doing, we
are operating the largest

13
vehicle fleet in Europe and
we started installing cameras,

14
dashcam, connected
dashcam in these vehicles.

15
So we are not only
moving parcels as we

16
are seeing, we are also moving data

17
while we are moving our parcels.

18
We are collecting data on the
streets every day, everywhere.

19
And what are the benefits of
choosing this type of vehicle?

20
I mean, obviously you come
from Geopost, so that helps.

21
Yeah, absolutely.

22
So we are relying on a very unique
asset, which is our delivery fleet.

23
Because of our core activity,
we go on all the roads,

24
any kind of roads, from rural
areas to urban areas every day.

25
So just to give you some figures,
we are able to cover up to 80% of

26
all the roads of the countries
we cover in less than 30 days.

27
And we are operating in more
than 22 countries in Europe,

28
including the largest one,
meaning we are collecting millions

29
hours of videos and kilometers
every month on the street.

30
No other player is in
capacity of doing the same.

31
So can you explain how that works
in the sense like you have

32
the cameras which are on the
vehicles and they are mapping

33
whilst they are there, they're
working while the van is moving.

34
What happens is the data
collected in the sensor and then

35
when they arrived at the depot,
let's call it depot, they,

36
it uploads the data or is
it sending data in real time?

37
And then also what
happens with that data?

38
Because obviously when it's imagery,
there's a bit of fear about it.

39
So tell me about that.

40
Yeah, and that's the beauty
of our system is we are

41
passively collecting data,
meaning that the camera

42
is only recording while
the vehicle is moving.

43
That's the first thing.

44
And the second thing, these
cameras are connected dashcams,

45
which means that they are
connected to the cellular network.

46
We don't need the van to go back
to the depot to upload the data.

47
We are continuously
uploading our telemetry data.

48
However, and that's the important
thing to remember, is we are storing

49
the images on the dashcam and we
only upload what our customer needs.

50
So we don't upload millions of
hours of video on the cloud.

51
We only upload what
our customer needs.

52
And this is also very important
from a GDPR perspective because we

53
are applying very strictly all the
GDPR principles for our solution.

54
We've seen computer vision solutions,
they pop up everywhere because

55
as we know there are many cameras
around cities, around everywhere.

56
Why is it better in your
opinion to have the camera

57
moving instead of maybe having
a solution that's fixed?

58
Well, that's a very
interesting part because as

59
you were mentioning,
computer vision is growing.

60
There is more capacities
with every kind of players,

61
but sometimes the lack of fresh data.

62
And what we are providing
is maybe the freshest

63
street-level imagery data in
Europe. It is incomparable.

64
We are the largest street-level
imagery platform in Europe right now.

65
So what our customers are
looking for is freshness,

66
is very, very recent data
on the street that will

67
help them understand what
is happening on the ground.

68
Like for example, a
smart city will wonder,

69
what is the state of my bus stops?

70
Can be this, can be
garbage, can be anything.

71
Thanks to our data,
they're able to get this

72
data very, very recently
in every street.

73
So this is like something
very new in the market.

74
Okay, and what are some of
the key use cases when it

75
comes to smart cities or
utilities that you have?

76
So we are addressing multiple use
cases depending on our customers.

77
We are addressing mapping, the
mapping industry, helping them to

78
list all the speed limits, the
traffic signs, the traffic lights.

79
But that's only a first part.

80
There is also road maintenance.

81
This is very interesting
for smart cities

82
to understand the
state of their roads.

83
It's better to detect
the potholes very early

84
than later because it
costs less money to fix.

85
But it can be way further than this.

86
You can be able to list
all your kind of assets in

87
your city thanks to our
data, from streetwalk,

88
from the stores, from— we are
talking about the garbage,

89
we are talking about
the parking spaces.

90
So it can be any kind of detections.

91
What about your platform?

92
What about what's behind?

93
Because I guess my question is what

94
differentiates you
from other companies?

95
So we wanted to have the most
scalable solution as possible,

96
meaning that we don't
have any interface.

97
It's just an API, a single API giving

98
you access to every country in Europe.

99
One single data flow,
one unique source

100
of data from one unique
model of cameras.

101
So you get very standardized
data thanks to our API.

102
So our API will only help
you to get the raw data.

103
Raw data, I just need to mention
that it's blurred for GDPR purposes.

104
We blur faces and license plates,
but you get all the images.

105
Up to the customers
then to analyze and

106
do its computer vision
over the images.

107
So that's how we operate.

108
We are a raw data provider.

109
And then how often do your
clients actually ask for data?

110
Because obviously you have
the vans driving around

111
constantly, but
realistically speaking,

112
how often do they need to
have that refresh of data?

113
So first of all, it depends on the use

114
case and the typology of customers.

115
But when it comes to mapping
players, usually they will need

116
to refresh their map once a
year with every speed limit.

117
So they will try, thanks to our data,

118
to map all the roads of a
single country in a year.

119
Okay, but for a city, for
example, it can be more often,

120
can be a monthly update depending on
the types of assets they are tracking.

121
So it really depends on their needs.

122
But the good thing is, as we
are mentioning, if you want to

123
have daily or even hourly
refreshed data, we can provide it.

124
It's really up to the customer needs.

125
Well, maybe also to mention that

126
our next plan also is
to go beyond Europe.

127
So right now we are covering
22 countries in Europe,

128
but we are now expanding
outside of Europe in Asia,

129
and there will be more news
to come very soon on that.

130
Okay, well, thank you very much.
Thanks so much for having me.