Subscribe
Share
Share
Embed
This is a podcast about the many phases of technology innovation: ideas, research, development, management, marketing, and many others. We’re talking to a wide range of people with diverse and rich backgrounds including all species of engineers, scientists, mathematicians, business people and technical managers. We hope you will enjoy these podcasts as much as we did making them.
Hello, everyone.
Welcome to the next episode of Hewlett
Packard Lab podcast from research
to reality.
This time, we are doing
the spotlight on animal tracking.
I have amazing pleasure and honor
to host two gentlemen here in the room
and one gentleman out of Africa.
let's start with a remote.
attendee.
Alex, would you mind introducing yourself?
Yes. Good evening to you.
Thank you for having me.
It's it's a great pleasure.
I, I met, HPE,
about two years ago
with the idea of creating this app
that would automatically be able
to identify animal tracks and, and prints
and, I'm happy to say
we are a long way down the track
and the signs are fantastic.
And it looks like
we are getting this right.
Thanks, Alex.
I'm sure you'll
be able to tell us much more about it.
Let's introduce, first. Shivang.
Hi. Shivang. Hi Dejan.
Nice to meet you again. yeah. Hi,
everyone.
I'm Shivang and I.
I'm a researcher at here
in the Networking and Distributed Systems
Lab working
with, Puneet Sharma, our director.
I've been here for about a year
and a half now.
before that, I was finishing up my PhD
at Northeastern University,
where I did my PhD in wireless networking.
Thanks, Shivang! Hi Lianjie!
Welcome back on the podcast.
It's a pleasure to be back, Dejan.
my name is Lianjie Cao.
I'm a senior researcher at Hewlett Packard Labs.
My journey with HP slash
HPE started with an internship
back in 2013, and I joined Labs
as a full time researcher
in 2018
after I got my PhD degree from Purdue.
Okay, excellent.
So I heard about this project
of animal tracking.
I always wanted to learn more.
I never had a chance.
So this is, you know, like,
a double opportunity both
to talk to you guys
and to learn a little bit more.
So who wants to,
take the first stab at it?
Yeah, I can start.
so, yeah, this project is basically,
quite an interesting project,
as you mentioned yourself.
so at a very high level, it's
basically about trying to digitize
and save this ancient art
that is called animal tracking.
And and what is animal tracking?
Basically, it's this ancient scale
that was evolved in Africa
thousands of years ago.
And basically what it involves
is these human trackers,
they go out into these national parks,
forest wildlife reserves,
they must form a mental image of what
the animals are doing,
where they're going, and that they don't
do that by befalling the animals.
They do that by looking at the signs
that the animals have left behind.
from that could be their pawprints.
That could be droppings
or other scratches that they leave behind.
and so that's what, animal tracking is.
And unfortunately, due to,
urbanization that's happening globally.
Right.
this ancient art
is sort of withering away.
And our goal here with Alex
is to try to save that ancient art.
Right. And we are building an end
to end system for that.
I'm sure we'll go into details, later,
but just briefly.
We have a front end.
We have a front end piece,
which is the mobile apps and the web apps
and then which help us gather data
and get images from the field.
And then we have our back end,
which has host and the intelligence.
Right.
The machine learning algorithms
and whatnot, to then take in
as input those images and then output
which species they belong to and so on.
So what was your first thought
when you heard about it?
Lianjie, about Alex and his project,
which really is amazing.
Shivang can start first because I
joined the project a little bit late.
I can add to that.
So yeah.
So, yeah, for me,
I remember Puneet one day just came to me,
and talked to me about this project,
and at first I was a little confused,
and, like, I didn't know what to expect
from this project.
but then, thinking in the whole sort of,
scope of the project,
I was very excited, basically, and trying
to see what all it entails.
And it's, you know, very different
from typical research projects
that at least I've taken or over
in my research career.
And, and also
when we we started speaking to Alex
for the first time,
it really clearly came across
how passionate he is about the project.
Right.
And that itself is very motivating. Right?
how passionate he is,
how inspiring he is.
And that helps
us push ourselves further too.
I mean, to
me, it's indeed a very different project,
a different project,
when compared to the self
entertaining kind of research project
that I do in labs.
and I joined the project
a little bit late,
mainly working on a machine
learning part of it.
and, but the first time
I heard about the project, like,
oh, man, finally we're shifting
our research directions
a little bit and starting to work on
some, something fun.
And I'm ready for an adventure
in the jungle.
Right. So. But the thing is, like,
of course, we never do that.
And, it's really a great and exciting,
project, like
Shivang said, and a terrific experience
to work with experts
like, like,
Alex from a completely different field.
Okay,
so we've been talking about Alex this.
Alex that. Alex.
How did you approach HPE?
Okay,
so I run an NGO, a not for profit company
in South Africa called the Tracker
Academy, and we train young rural people
from big wildlife areas
in traditional skills of animal tracking.
A one-year formally accredited program,
we are a school, and then we deploy
our graduates into conservation jobs.
And it's just become evident
over the last few years
how the ancient traditional skills
of tracking in Africa have diminished.
We are losing these skills and waste.
And at about the same time
I met with, Dobias van Ingen from HPE, he,
from Aruba,
and he was in South Africa on a conference
and we happened to meet
just coincidentally.
And I told him about the idea
that we needed
to digitize this ancient data
before we lose it completely.
And and he took the idea back to HPE.
And that was two years ago.
And here we sit now with,
I think it's version three of the app.
and we are a long way down the road.
And I must say, there's been great success
and it's been such a pleasure to work
with such an incredibly talented HPE team
and the HPE labs.
What these people have been able to
achieve is just astonishing.
So I can
thank you Alex I can hear
all this excitement from your side,
but I'm curious from far away Africa,
how did it excite you here in Milpitas?
Lianjie and Shivang. Sure.
so I think to me it's the impact.
So we have been working on
an innovative idea.
like challenging problems
in the labs, but those ideas and those,
challenging problems
usually take a very long time
to see the impact, like, for years.
But this project is very different
in the sense
that we can directly, almost directly
see the impact immediately.
And all the improvements and
all the enhancement we made in the system
are going to be used by Alex and his team
on the field, like right away.
So there's no way do you see the impact?
And I think also the impact of sort of
like a beyond labs, even beyond HPE.
Like we have been always saying
accelerating the impact,
being a force for good.
Yeah.
And I think, to add to what
Lianjie said,
he mentioned a bit about the social
impact.
But, I mean, even from the technical side,
as researchers,
right, we get excited about tackling
challenging problems.
And this, again, I did not
you know, we have
we should expect challenging these sort
of challenging problems at the beginning.
But as we went through the project,
we encountered several challenges
and technical challenges
that we had to overcome.
And as researchers,
nothing more exciting than that, right?
Yeah.
So I can see why historically
this skill is very important.
But how is it still important today? Alex.
Yes, that's a good question.
Today, in modern day conservation efforts,
we need trackers to track to collect data
on behavior of animals
and the trackers,
but possess the knowledge
able to to accurately identify
and interpret
the signs of the animals left by animals
and thereby able
to collect important data that otherwise
would not have been collected.
that trackers are able to know
what animals are doing
even when they're not there, by
by following and interpreting their signs.
We also need trackers in anti-poaching.
I'm not sure if you know,
but there's been a crisis
in southern Africa of rhino poaching.
We've lost something
like 70% of our rhino population
in southern Africa over the last decade
and a half, and we need talented
trackers who can pursue the poachers
on foot and track and find them.
So it's,
the trackers are the unsung heroes,
and their skills are becoming
increasingly important
as as we go through time
and as conservation management
start to realize the value,
the tangible value
that this ancient skill brings,
the modern day conservation efforts.
So I can understand
people in the field,
and all the challenges they have.
I can imagine them.
But what are the challenges for
you here in Milpitas?
Shivang and Lianjie,
I can start, like.
So our goal, the HPE team is to build
a digitized solution
for Alex to solve all the challenges.
he mentioned.
And the the challenges for building such
systems is, like, every step of the way.
So if you think of this,
like starting with,
we need to figure out the design
of the architecture
for this end to end system,
from capturing the footprint in the jungle
to infer the images of the species,
in the back end.
Right.
So, for instance, we need a front end
to capture the footprints and annotate
the footprint with the information
related information before they get lost.
And we also need to upload them
reliably, to the back end.
So all those kind of things
we need to take care of even in the back end,
we also need to think about
how should we clean the image,
how should we store them, label them
and then infer the
the species of the footprint,
all those kind of stuff?
Yeah.
And to add to
that, I mean, we have way more challenges
because of the fact that
the geography of the, the problem itself.
Right.
A lot of these trackers are working
in remote areas where there is like
little to no connectivity at all. Right.
And we are asking them
to upload these images from those areas.
So how do they do that?
There are all sorts
of technical challenges that arise
because of that lack of connectivity.
Are there any non-technical challenges?
Alex.
Because here we only jump on technology.
Technology.
I'm sure there are other kinds of problems
that you are,
exposed to on a daily basis.
Well, yes.
In in training the machine, the algorithm,
we need to collect many, many examples.
Let's take for example, we want to train
the machine to identify a lions track.
We need to to, to up to identify,
photograph and upload at least a thousand
different examples of lion tracks.
And so that takes time.
And you so and it takes people
that who know how to identify
a lion track easily.
If we have less than competent trackers
training the machine,
it's, it's going to have a great impact
on the on
the outcome of the,
the the performance of of the app.
So we've got to make sure we've got expert
trackers out in the field collecting
literally thousands and thousands
of images of 120 different species.
So that's an incredible feat.
And it's it takes time
and weather often is a problem.
so yes, that's that would be the most,
apparent challenge we are dealing with,
but we're getting through it slowly.
I think we've, collected
about 30,000 images so far.
So you gave me
some hint that you are using an app
to solve the problems.
Can you tell me how you are
solving the problems using this app?
sure.
So, for the app side, we basically have
like two different types of apps.
We have the mobile app and installed
on the smartphone of the trackers.
And we also have the web app.
So for the mobile app,
the users can create
different type of account like public account
and professional account and things like that.
And then in their account they can browse
first all the images you have taken
before with the different types
of information embedded.
And they can also, of course
take new photos of the footprints
and then at the same time
associate the metadata information
of the of the species
and things like that.
And then and finally they can upload
your images to the back end.
And in the web app, the app,
the web app is totally different.
It's more for the purpose
of analysis and management.
So in the web apps,
the administrator can go ahead
and browse all the images uploaded
by different users.
Professional, public
and all those kind of stuff.
And they can check
into the detailed information of them
and try to, modify
the information of some of them are wrong.
And what's more interesting is
they can also browse the images,
filter them and show them or, you know,
the active map to see the geographical,
you know, information on the map
so they can get a better visual
understanding
of where the animals are spotted
and all this kind of stuff.
And in the future,
we are also thinking of doing more deep,
deeper kind of like analysis
and understanding based on this type
of geographical information.
For instance, we can try to learn
how global warming impacts
the animal migration at a very high level
and things like that,
but also like to highlight the fact that,
I mean, the challenges that we describe
this, such a diverse set of challenges.
And so we needed a diverse set of people
with diverse skill sets
to tackle those problems. Right.
So this effort is sort of a pan-HPE
effort.
Over the time we've gotten more
and more collaborators from outside labs.
So for example,
we have Ezmeral BU
who are providing us with the back end
to help us run the modeling
and the inference on, then we have,
the Asia Pacific Innovation Center team.
Who is
Who's helping us
build the apps themselves. Right.
Because, I mean,
these apps need to be very professional.
Grade.
We can build some apps,
but I'm sure they won't be
as good as what these guys put out.
I mean, these guys are really talented
at what they do.
so yeah, that's
just something I want to highlight
I'm glad you started to talk about business.
you have so far
received help from various businesses
like Ezmeral or Asia-Pacific teams.
But in the big picture,
how do you see this becoming sustainable?
Not necessarily making money out of it,
but someone needs,
on a daily basis, annual basis
to support this.
So it should eventually go
into some business, into some production.
How do you gentlemen see this?
so I think, the idea who we are
at the moment, currently we are relying
more on the goodwill of, let's say,
HPE and Alex that, we can sustain
this project, in the short term.
But I think in the long term,
it can also be about once we
if you want to monetarily sustain
it can also be about
how we gamify the app a little bit,
maybe for like folks like you and me.
Right.
Like we can go out in the field, snap,
you have pictures of footprints
and send it up and get a result.
Right? Okay. This is a lion’s footprint.
This is a wolf’s footprint.
And then maybe that builds
this sort of ecosystem of users
that, the an end goal could be.
And then that could ultimately help us
sustain, what we're doing.
And hopefully, I mean,
we always will rely as, again, HPE's
nature of being a force for good, right?
I mean, that's always a thing
that will hopefully help us
sustain this in the long term. Yeah.
And to add a little bit to that.
So this is becoming something that the
the impact of this project is going beyond
HPE because we have been talking to
and Alex as well, talking to some other
service providers,
they are willing to provide
like hosting services to the project
because you want to advertise, you know,
of course,
they're on those and things like that.
So those can become a more,
you know, general coverage team
across companies and things like that
and be a very good showcase of HPE's
technology.
It can be,
can be applied to solve such problems.
So we can get some,
you know, some similar attentions
and projects from potential customers.
Any opportunity for our Aruba
because they are our edge side.
Yeah.
So yeah, as I mentioned briefly before.
Right.
the work that a lot of
these trackers are doing are in
remote areas where there is
basically no connectivity at all.
Right.
so Aruba,
basically us working together with Aruba
can help us tackle
that specific part of this problem where,
maybe using technologies
like satellite or private 5G,
we can have some deployments
in these remote areas.
that can help get connectivity
to those folks that don't have it.
Right.
And that will ultimately help us reduce
the latency that these guys face in terms
of uploading the images, getting results
back from the servers and so on.
And if we can do that, obviously
that will ultimately help the quality
of experience that these guys, face
in these remote areas.
So accomplishing what you have,
which is really outstanding,
what was the most challenging about it,
and was there anything
unusual in terms of your solutions
or surprising things?
I think to me the most challenging part
is actually labeling the images,
which is kind of like strange
because Alex came to us for help,
and it turns out that we need his help
to label the images first
before we can apply technologies
like cloud computing
and machine learning to help him.
So it's it becomes some sort of
like a chicken egg problem.
So. Well, you want me to help you,
then you need to help me first.
That kind of stuff.
Which is kind of surprising to me.
and if you think about this, like, what's
worst in this case, in this problem, is
we there is nobody
else except Alex and his team
can help us to solve this problem.
Because these guys are professional for me.
Like, I can tell
the difference between cats and dogs,
but I cannot tell the difference
from the footprint. Right.
So and this is also an in from the other,
perspective, it's like,
there are always something we take for
granted in our research project,
but in reality, those kind of stuff
like a label that you measure
and things like that
are probably not quite there
And, Alex.
Have you considered any standardization?
I mean, there are different
classes of animal tracks and all of that.
Yes. Well, there's standardization
in terms of trackers skill set.
we are able to objectively evaluate
a tracker’s
skill set in the various components,
practical components of tracking.
And that is what is established.
And,
the South African Department of Education
recognizes, our standards in that regard.
And there are standards
in the US as well, run
by other companies that that promote
and do tracking assessments.
But in terms of, standardizing the data,
that's very hard.
And that's I just wanted to add
the most complex part of this program,
this whole project
is that a lion footprint.
No two lions footprints are the same.
And it depends on the type of soil
that it stands in
or how it's moving,
if it's turning, if it's running,
if it's old or young, it's
constantly throwing up different types
of, of of forms
of, of the same species track.
And that's why we have to,
get, get so many,
capture so much data
to be able to train, train the machine.
So that is, that is,
that is the most difficult part of it.
And you know, we there there are similar
apps out there that identify, identify
plants and trees by taking a photograph
of the leaf or the flower.
That's a that's a far
simpler, procedure and project
because there's, there's much less
very variation between an apples
leaf that grows in the north of the US
and then in Europe,
whereas in, in, within tracks, as I said,
there's just a great variability.
And it
seems intuitive to me that this is
very ethical, what you are doing.
But have you ever considered
that angle of ethics, of animal tracking?
Yeah.
I mean, at the Tracker Academy
we have a whole module on ethics
because many of our graduates
go into protected
areas, government protected areas,
and have to protect endangered species.
Excuse me.
And so they have to
they have to undergo polygraph
testing every, every six months.
and their livelihood
depends on being ethical.
and also from from another
completely different standpoint,
the, the,
the observation of an animal's track
or sign
and thereby being able to interpret
what the animal is doing
is a completely noninvasive, approach.
You don't have to go off to the animal
and dot it
and put a collar around it and caught it
and give it drugs.
Trackers are able to tell
where animals are moving
and what they're doing, as I said earlier,
without even having to see them.
And and that's very much goes
to this idea of sustainability,
of low impact on the animals,
not disturbing them.
And I think that that certainly falls
under the, under
the umbrella of, of ethical conduct.
Thank you.
Alex.
Coming back to Milpitas from Africa,
what did you two gentlemen find
most rewarding in pursuing this project,
other than being extremely exciting?
Yeah, I can start. yep.
So for me,
I think, as you mentioned a few times
already, it's just the direct impact
that is there of the project.
I mean, all the projects we do,
I mean, we would like to think
that they ultimately help society
in some way or the other down the line.
Right.
But with this project, you can really see
the directness of the impact.
Like whatever we do,
it goes into the hands of these trackers
that are going out into the wild.
and ultimately been the goal, obviously,
is to contribute
to the well-being, of our planet.
Right. Finally.
That's what, we care about.
So that's really the most rewarding part
for me.
Yeah, to me, like,
I learned how to track animals, of course.
And the other thing
I also realized through the project,
it's basically the gap between research
project and real world
problems.
For instance, the problems
that Shivang said, some of them are probably not
technically difficult to solve,
but we do need to, to think from user's
perspective wearing their shoes to
identify and to realize the problem before
we can do something about it, though.
That's the thing I learned. And, Alex.
And, Alex, why, does the audience
why does the public care about this?
Now? That's a good question.
The public should care about it
for a few reasons.
One, it is part of ancient
African cultural heritage.
and it has it
certainly in South Africa, due
to the effects of the apartheid regime
and, and forcing people off their land,
as well as the
the rapid worldwide trend of urbanization,
people have forgotten
how to and lost the skills of tracking.
And what I have seen
is that when we engage rural people,
many of whom are unemployed,
living on the outskirts of these big
wildlife areas, some of the last
viable wildlife areas left on the planet,
and they stare jealously
through the high wire fences.
One way to capture the hearts
and minds of these people
is through their own traditions, that
being the traditional skills of tracking.
And so it has a twofold function for me.
it's it's a skill
that can be used in a, in a technical
manner
to better protect, endangered species.
But it's also a skill that goes directly
to the hearts and minds of the people
who are the custodians of these wild areas
and these,
and these endangered species.
So that's why I think the public
should be caring about this project.
Great explanation.
So we came close to the end,
of this discussion.
Can we just close by,
each one of you very briefly stating,
what do you do
when you don't track animals?
Yeah, I can start like, I basically do
hikes, like a long hikes like five miles.
Ten miles.
I play badminton from time to time.
Used to play more often, but much less.
Now I play soccer sometimes I
also do a little bit woodworking at home.
I have a small garage shop kind of stuff.
Nice. Yeah.
Thank you. Yeah. For me, I'm.
I'm getting a big sports guy,
so I like to play a lot of sports, like,
tennis, badminton, cricket.
I don't know if people know about,
but it's a great sport.
it, and, I seen with,
watching a lot of these sports as well.
Then I like to attend, concerts
of artists, music artists that I like,
like watching stand up comedy.
That's authentic.
Thanks. What about you, Alex?
I love I love Shivang's answer.
I have two daughters
they are six years old and eight years old.
They were both born at Londolozi Game
Reserve, where I worked for 23 years.
we've now moved to the town
because of, their schooling.
We needed a better school for them.
But what I do do a lot of is
take them into wild areas,
and I'm trying to introduce them.
Well, I'm not trying, but I'm introducing
them to to tracking and tracking.
It's just such a wonderful way
to get people to immerse themselves
in wildlife,
to become connected with nature.
And I'm seeing the vehicle of tracking,
doing just that.
And I love spending time in nature
with my two little daughters.
Thank you very much, Alex.
Virtual handshake to you
and handshake here to Shivang and Lianjie.
I really enjoyed this.
I'm sure our audience will too.