October 17, 2023 • 35 mins
AI can solve some of today's most complex challenges, and over the years this has become reality even in the agricultural industry. Due to environmental factors and other threats, sustainable farming is becoming more at risk, and by harnessing the power of AI, tools to help local farmers are more accessible on a global scale. In this episode, learn how Rishikesh Amit Nayaka and Niharika Haridas used AI and Intel’s OpenVino technology to detect pests, and make farming equitable and successful in India. Additionally, they are joined by Intel’s Director of Government Partnerships and Initiatives for Japan and the Asian Pacific, Shweta Khurana, who explains Intel’s work with developing the latest voices in AI innovation.
Learn more about how Intel is leading the charge in the AI Revolution at Intel.com/stories
See omnystudio.com/listener for privacy information.
Mark as Played
Transcript
Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
Speaker 1 (00:03):
When a lot of us think of farming, it reminds
us of simpler times, and perhaps it feels like one
of the remaining industries exempt from the influences of the
modern tech world. But imagine a world where the success
of your family's farm crop yield is access to AI tools.
There's so much labor and effort that goes into maintaining
a farm, especially when farmers have to anticipate unpredictable weather
(00:25):
patterns and unprecedented seasons brought on by climate change. Plants,
like humans, are living things, with millions of tiny organisms
both attacking and assisting their life cycle. Some threats to
crop life are smaller than the human eye can see,
and when not addressed, the results can be disastrous to
local economies. But what if AI could solve the problem.
(00:46):
Giving eyes and access to where farmers cannot reach. AI
protects crops and the economy from the threat of microbial pests,
resulting in a more prosperous tomorrow. Hey there, I'm gram
Class and this is technically speaking an Intel podcast. The
show is dedicated to highlighting the way technology is revolutionizing
(01:08):
the way we live, work, and move. In every episode,
we'll connect with innovators in areas like artificial intelligence to
better understand the human centered technology they've developed. There has
always been a disconnect between nature and technology. However, today
there's a lot of science and technology at the core
of modern farming, and we're not talking about GMOs. One
(01:29):
of the biggest issues in agriculture is environmental threats. Farmers
struggle with protecting crops from diseases and pests without using
tools that could adversely affect consumers. AI has been instrumental
in helping farmers detect pests before infestations occur and result
in huge crop loss. But before we get into exactly
how it all works, I want to introduce our guests.
(01:53):
In twenty seventeen, Rishi kish amitt Nayak's family farm in
India so ninety percent crop loss due to pest in infestation.
After partnering with a fellow student, Niharika Haridas, the two Megatronics,
Robotics and automation engineering students found a way to use
AI to develop a method that could detect crop pests
(02:13):
through thermal imaging. This system, called kishan No, has been
proven effective and very affordable to local farmers. Rishikish America
thanks for being here.
Speaker 2 (02:23):
Such a pleasure to be here, Graham, thank you for
the invitation.
Speaker 1 (02:26):
We're also joined by Shwita Karuna, intel's Director of Government
Partnerships and Initiatives for Japan and the Asia Pacific. Sharita
has over twenty three years of experience creating trusted government
relationships and fostering government programs that encouraged the implementation of
modern science into the workforce PLUSH. She was instrumental in
(02:46):
helping kishan No grow as a farming tactic across the region. Welcome, Shwrita.
Speaker 3 (02:51):
Thank you, Graham, such a pleasure being here.
Speaker 1 (02:56):
So let's start at the beginning a very interesting story
around rishikish Can you tell a little bit about the
problem that your family and other farmers experienced back in
twenty seventeen.
Speaker 4 (03:07):
In India particularly, it's an agricultural country, so more than
seventy percent of the people do agriculture as their own
major occupation. In twenty seventeen, my father's grandfather was completely
invested into agricultural farming, and during that time, in Orisa particularly,
there was a plant best attack that couldn't be identified
for a longer period of time, and that resulted in
(03:28):
a lot of crop losses and hectares of land was
just lost because of an unidentified pist. Personally, we saw
a lot of farmer suicides in our own village, and
that was the major reason when I thought, Okay, I
do have a background of engineering, I do have a
background of robotics, so why not to create something for
our own farmers. And being part of that family where
(03:50):
we do farming in our parental site, I was just
touched with that fact that I need to do something
for the farmers.
Speaker 1 (03:58):
In Rishikishi's village alone, there were four farmers who took
their lives as a result of the devastated crop, and
his family saw a ninety percent crop loss that year.
The infestation was so devastating to their livelihood his family
considered leaving farming all together. And to make matters worse,
the problem was difficult to identify and trace. Before we
(04:23):
get into the actual details of how you solved it
in Arika, how did you get involved in the project.
Speaker 2 (04:30):
I decided to pursue mecatronics and automation at Viatchene out
of a sheer passion for robotics as a twelfth grader.
So I came across the work that many companies like
Boston Dynamics were doing at that point and exactly right
the spot pro vote of course, and I was just
enthralled with potential that it helped, Like it was like,
(04:50):
oh my, what this could change humanity?
Speaker 5 (04:52):
And I was like, I need.
Speaker 2 (04:54):
To do something in this space. I wanted to help
people with this new technology. And that's how I went
to Aitchen and that's where I'm Metri Shikish and we
started talking and we were talking about this project and
I was like, you know, that's that's amazing that we'll
let me contribute to it as well, and that's how
we started collaborating on the project and then we participated
(05:14):
in the Inaugril Intelliet Global Impact Festival and the rest
is history. We had a wonderful time and you know,
the support that we have gotten from Intel for it
as well has been phenomenal and that's the reason that
Kishano is at the place where it is right now.
Speaker 1 (05:29):
Excellent. So now as the I guess the sixty four
thousand dollars question is how does the kishan No work.
Speaker 4 (05:37):
Kishano basically taps into saturate based thermal imagery. These images
can detect temperature variations and crops which often indicate Streuss
disease or pestal activity. For instance, areas affected by certain
pest or microbol infestations may exhibit different thermal patterns compared
to healthy areas. We collect images from Sentinel two and
lands At eight satellites. Those satellite images are then sys
(06:00):
to get index mapping out likes, for example, vegetative indexes
and moisture indexes through a software called QGIS, so it
basically gives us the values for those vegetative indexes and
moisture indexes, and these gathered thermal imageries processed using AA algorithms,
where we've processed the images first into the open Veno
platform and we get a d blood image for better
(06:23):
accuracy of training of the models. Then these algorithms are
trained to recognize patterns or animalies that correspond to microbilan
pest outbreaks. Over time, has more data is collected and analyzed,
the AA model becomes more accurate and efficient in its
prediction and leveraging the power of machine learning. Once a
potential threat is identified in the system, the systems can
(06:43):
send alerts or recommendations to the farmers in the local
administrative levels, where we also design the physical device apart
from the AA algorithm to get a confirmatory test that
there is a pest or plant disease outbreak. This actually
includes information about the type of threat, it's severe, and
recommendation algorithms.
Speaker 5 (07:02):
This proactive approach.
Speaker 4 (07:03):
Helps farmers to take actions before the problem becomes widespread
and saving both time and resources.
Speaker 1 (07:10):
I'd like to talk about Intel open Veno a little
bit so quickly, just to inform our audience. INTE open
Vino is a cross platform toolkit developed by Intel that
deploys deep learning models on visual data sets, helping computers
better recognize and process images to inform decision making. But
I'm curious as someone who's just as interested in what
(07:30):
didn't work as opposed to what ultimately does. Why did
you decide to use Intel open Veno. Were there are
other methods you tried first?
Speaker 2 (07:38):
So we did try a lot of techniques, and we
found that open Veno worked perfectly with our project, especially
with the goal that we were trying to achieve. So
we saw that the hardware requirements as well as the
software requirements did completely match. Also, we had mentorship from
Intel and we were able to properly and in a
better way adapt to those systems to our project, and
(08:01):
that's the reason which it was open.
Speaker 4 (08:03):
We know, we actually tried to degler images through some
deep learning algorithms, but those algorithms was actually not satisfying
the accuracy that we actually wanted, so open Veno just
suited out the case perfectly.
Speaker 1 (08:17):
One thing I'm interested in is the pests that were
being detected. Am I right in saying that it had
a unique therm signature?
Speaker 5 (08:24):
Yeah?
Speaker 1 (08:24):
And how did you discover that?
Speaker 4 (08:27):
In twenty seventeen, Once we identified the problem, we actually
tried to create a physical device through a thermal camera
set up and microprocesses. We were rotating that device among
the periphery of the crop fields to understand what exactly
the thermal traces are in the leaf of the crop plants.
And once we understood what are the thermal signatures for
(08:47):
different crop plants, we understood there is a concept that
whenever there is a pathogen or a plant disease, there
is a certain increase in the leaf temperature. And if
we identify that leaf temperature increases in the particular or
in a particular duration of time, we can actually significantly
say that there is a best attack or a plant
disease in the crop area. Once we had the theory,
(09:10):
we tried to incorporate that similar formula in the vegetative
index of the satellite setup. So in twenty nineteen we
had the physical setup, we tried the same literature to
understand it to the satellites.
Speaker 1 (09:22):
Hearing Rishikish and Aharika elaborate on how they design their
imaging tool reminded me of my own experience attempting to
develop systems to work remotely in the jungles of Africa.
It's not an easy feat though, as there's no real
infrastructure for these sorts of products, especially when they are
limited by internet access and availability in the area. Hearing
(09:44):
how much progress these two had made with their program,
maybe wonder about the challenges that went into making this
tool available in the rural farmlands of India.
Speaker 2 (09:57):
There has always been a digital divide in India, as
we can see, but now it's been narrowing and that's
a very good news for all of us, and that
infrastructure is also becoming better. There's also research that India
has one of the cheapest internet out there in the world,
so I mean, it's being adapted and we are glad
that it is. But when we were working on it,
(10:17):
we did face a lot of infrastructure issues regarding internet
services as well and internet connectivity exactly, and sort of
having that satellite imagery. Gaining access to the satellite imagery
was very difficult for us because that area wasn't mapped.
Remote areas aren't usually mapped with that much precision as
that of let's say, an urban area, so we did
(10:38):
have some issues with that, but then we did try
our best to solve those and gain satellite images from
the areas that we neated.
Speaker 4 (10:46):
Farmers in the villages particularly, they were quite a bit
skeptical to try this out, and the farms because in
India particularly didn't back that time, we didn't have that
much of agritechnology tools or products, and going as a
youngster something around in class ninth or tenth and trying
out as some different new projects or new census in
(11:07):
the field, they were quite a bit skeptical. So managing
that side of that, Okay, we are doing something good,
we are doing something better for your own crops, we
are doing something for the best of the society. Convincing
them was one of the very huge challenge over there
in India.
Speaker 1 (11:23):
What kind of data or training processes were involved in
training the model to recognize microbiopests in the crops.
Speaker 5 (11:31):
Initially it was only deep learning algorithms.
Speaker 4 (11:33):
Further on, when we had a lot of thermal praise
data and we had did the d blood images, we
were just focused on the CNN models to train the data.
And it hadn't given a good accuracy of for around
ninety points something percentage, so it was a pretty good
accurate to start with for a particular set of crops.
Speaker 1 (11:51):
You said, CNN, could you just define what that is please?
Speaker 5 (11:54):
Conventional neural network.
Speaker 1 (11:56):
Okay, And that's just another AI technique to for learning.
Speaker 5 (12:01):
Yeah, yes, a machine learning okay, okay.
Speaker 1 (12:04):
And you just mentioned about the accuracy that you achieved.
Would you say that's typical for the Intel Open Veno
platform to get that sort of result.
Speaker 4 (12:14):
The accuracy is for the total accuracy of the model
for a particular set of crops, for example, tomatoes and wheat.
For those two crops we had an accuracy fround ninety
point two eight percentage, and for other crops it's still
in the process of getting more accurate and all. So
for these two crops, overly, it was the accuracy that
we measured out and.
Speaker 1 (12:33):
In terms of the Intel Open Veno technology, can you
think of anything any other farming use cases beyond pest
management and crop protection.
Speaker 5 (12:42):
Currently, we were trying to work on crop genome analysis
where we were actually trying to understand because of the
climate change to the new variants of crops are needed
to adapt to the new climatic conditions. So we were
trying to understand how exactly we can use machine learning
algorithms to create new genomes in the crops the microbiology
side of it.
Speaker 4 (13:02):
So yeah, that's one area that I was completely focused
on in this past recent days.
Speaker 2 (13:08):
I would like to add on to that, and as
Education mentioned, convolutional neural network model that we used, it
was at that point not something that was used by
the AI community, but then we now see a lot
of use cases for that and that's something that we
are very glad about. And also some of the use
cases that I have at least found as an AI
enthusiast that models like these could have is in real
(13:32):
time data, especially as the climatic change has become a
huge issue. It is something that can help a lot
of farmers with when there is excessive rains or when
there is no rain at all, to predict these through
AIML technologies. And I believe that the limit is boundless
when it comes to AI technologies. Right we are seeing
a start of a new era of AI, and I
(13:54):
am very glad to see how I was being used
by lots of companies, and we also hope to go
contribute to that, and I hope for a very bright future.
Speaker 1 (14:06):
AI has been the focus of a lot of discourse
over the last couple of decades. While many of us
experience it as virtual assistance in our phones and computers,
AI has been giving us listening, watching, and reading recommendations
for years and we continue to see it evolve and
even create content like images and written stories. But that's
(14:28):
all just the beginning. AI has so much potential to
positively impact the way we work and live. It can
be used to detect new variants and threats in agriculture
brought on by climate change conditions. The Intel Open Vino
technology played an essential role in this, providing higher accuracy
for detection. I'd just like to switch a little bit
(14:51):
to the agribusiness side of things. And maybe I can
get Shuita to comment on this in terms of the
Intel Open Vino and its app cation here for pest detection.
Do you see it complementing other pest control methods in
agriculture and does it have the potential to replace pesticides
and insecticides and farming replace.
Speaker 5 (15:13):
Is a little on the harsher terms.
Speaker 3 (15:16):
What I would actually look at it is AI and
agricultures really helping farmers make data driven decisions, optimize crop
yields conserved resources like water and energy. The challenge here
is not just the solution part of it is also
kind of encouraging next generation technologists student innovators to come together,
look at the local problems like what Neharikan risikation have done,
(15:40):
and then create a solution using all the skills they've
learned as part of their formal education as well as
as part of being a part of Intel programs the
Interdigital Rediness Program portfolio, come together and democratize AI skills
in a way which gets a common person a farmer,
to understand trust and emergingology like artificial intelligence and hopefully
(16:02):
become comfortable in applying it to solve the daily problems
they would be facing as part of their community.
Speaker 1 (16:09):
I love that term democratization of technology, and I think
that's ultimately what technology does is get it more accessible
and cheaper to get it to the far regions of
the world. I'd just like to expand a little bit more,
maybe if you could explain some of the programs that
are available through inter Open VENO to help farmers or
(16:29):
businesses with limited resources to get access to this sort
of technology and expertise.
Speaker 3 (16:35):
I'll just take a step back here, right because we
keep talking about increasing digitization, which today a lot of
governments and countries are going towards. But what it really
means is when we focus on increased digitization, we also
need to invest more in building the digital readiness of people,
specifically in terms of emerging in critical technologies like AI
(16:55):
or what you spoke about, like the usage of open Wino.
How do we get person to understand how they can
utilize the technology like open we know to be able
to solve their local problem and create indigender solutions. So
all this kind of comes together through a whole program
portfolio which we have which is called the Intel Digital
Readiness Programs, which is driven in collaboration with government, academia,
(17:19):
civil society, and the industry and focuses around building shared value,
shared responsibility so that we can really demystify democratize these
superpowers which we keep talking about, like artificial intelligence for
a very broader and a diverse audience for young budding
technologists like Neiharika Ushikish but also for those who are
(17:42):
going to be consuming the technology at the other end
of the spectrum. The programs are a lot, they're many.
They range from you know, programs like AI for Citizens,
which talks about getting a citizen to understand how to
navigate his or her life in an AI driven world.
AI for Youth, which really allows us to empower youth
with not just the technical skills associated with AI, but
(18:05):
also the social skills in a very inclusive manner. And
then we have AI for Future Workforce, which is for
vocational community college students, engineering students, which really helps them
to understand how to be prepare themselves for becoming a
part of the future workforce. So a huge spectrum, lots
of programs, but the one which is very special to
all three of us in this case, and I'm sure
(18:27):
Education Aherka would agree with that is our EI Global
Impact Festival, because this is where we work with all
these student innovators. We get them together and we get
them to celebrate their AI innovations with a huge passage
which does not just allow them to showcase what they've built,
but also helps them hone their skills by getting mentored
(18:48):
by Intel technologists because at the end of the day,
these young students are the next generation technologists, so we
want to make sure we work for them to support
and build an AI ready generation.
Speaker 2 (19:00):
Platforms like these have been really instrumental and I have
seen the impact on ground that they make in supporting technologists,
young technologists like us, and we have always been very
grateful for the opportunities and mentorship as well that Intel
has provided. And that's something that we wish that every
budding technologist in India and all over the globe can
(19:22):
at least experience, because mentorship and guidance is an important
pillar of one's journey and having someone who can teach
you more about AI, how to use AI, and how
to benefit from AI, especially with the immense potential it
has that is life changing.
Speaker 1 (19:42):
You're listening to technically speaking, an Intel podcast will be
right back. Welcome back to technically speaking an Intel podcast
(20:03):
shweeta last episode of this podcast, we talked with Reachhuvu,
one of your colleagues, about the ethics and responsibility of AIM,
wondering if we could get your thoughts on how you're
working with governments and industry leaders around AI and trying
to help them navigate some of the ethics and responsibilities
(20:24):
around AI development.
Speaker 3 (20:26):
That's a very interesting question for us, right because when
we speak about digital reddiness or how do we build
digital readiness, we look at three pillars. Largely, one is,
of course learning the skills of emerging technologies like AI,
but more importantly, getting to understand and trust those skills,
So getting to understand not just what the advantages are,
(20:47):
but also what the pitfalls are. Getting to understand which
situation should we apply the emerging technology in and which
ones we should abstain from. So our programs, in fact,
inculcate a lot of discussions around these there is, which
range from the ethics piece of it, which range from
how how do we make it more inclusive, how do
we make it more diverse? And so much so that
(21:10):
if you kind of package it all together, it comes
under the larger umbrella of responsible AI. So how do
we really encourage not just youth, but every citizen, which
includes the governments who we collaborate with and partner with
to understand what is the responsible use of these superpowers
like AI to gain broader socioeconomic benefits for everybody.
Speaker 2 (21:32):
As a youth igffellow. That is exactly what I focus
on Internet governance right and how AI governance works and
how AI can be regulated. But then what about AI innovation?
It shouldn't be regulated or stifled due to laws coming
into place that can have that effect where people continuate
and they can't contribute to new technologies, so that there's
(21:55):
a delicate balance between them, and that is what I
also do look into. And the whole area of how
becoming emerging technology is like even robotics which has a
huge artificient intelligence ethics background out there, So how do
we harness this without harming humanity? And that is something
that I believe all stakeholders, including the youth of our
(22:16):
country or the globe, should be focusing on because there
also tends to be the whole bias of youth not
being given a voice when it comes to these emerging technologies.
But I believe if they do understand what it is
about and what potential risks it has and what potential
benefits it has, that gives them the knowledge to use
it responsibly and ethically.
Speaker 1 (22:39):
Using AI can be as complicated as Niharika has pointed out,
but the tool she and Wishikish have been able to
create from this place of innovation and AI have changed
the world for the better and they have the results
to prove it. In terms of the Kisheno technology that
you have developed, do you have any stats on the
(23:02):
crop that has been saved or the reduction in crop loss?
Speaker 4 (23:06):
In twenty to twenty, we actually piloted this around in
eight districts in Orissa and more than around seventy two villages.
We actually serve it upon and piloted upon and for
one season we tried it particularly on wheats and tomatoes.
Once we had data that we could actually predict that
there is a pest attack or plant this is coming up,
(23:27):
we use that data to try to save those fifty villages.
We used pesticides and fertilizers just before whenever the pest
and pest attack could have happened, So it actually saved
around those fifty villages.
Speaker 1 (23:40):
I'm really interested in how the technology actually is deployed
and distributed to as many villages as possible. To me,
the innovation is part of that as well. How do
you deploy it, how do you scale it? And you
said you went to seventy two villages, how did you
get to all of them and provide this service and
this knowledge.
Speaker 5 (24:00):
In the local districts.
Speaker 4 (24:01):
We contacted the administrations and with the recognitions we had
with until it was really easy to contact the administrations.
So once we had contacted the administration the local villagers, they
were actually understood, Okay, there is someone who is coming
to do something in their villages and it won't harm them,
So they were at least a relaxed that nothing is
going to be happening.
Speaker 5 (24:20):
And also they actually co operated out.
Speaker 4 (24:22):
So we had to draw the plots, We had to
map it on the satellite software that we had and
it would actually give us a satellite based crop image.
And for each crop images, we just needed to market
around the perimeters of that particular individual farmer and the
work is done. We just needed to understand how what
area that particular farmer has, what is the crop type?
(24:44):
When did so what is the raining patterns and what
is the soil type. With these certain parameters understood, the
farmer had to do nothing. We were sitting on a
room played server and we were training these images and
it was again the process kept on going. We had
the results each week, we just to share them. Okay,
this is the condition, this is what your crop health is,
and your crop is safe and if not, we will
(25:07):
at least give them some predictions.
Speaker 2 (25:09):
One of the other plus points or advantages of our
innovation was how cost effective it was. So now this
is a huge issue when it comes to India that
technologies are out there, but they can be very expensive
and that's not reachable to a conventional Indian farmer. They
need solutions that are cost effective because of budget constraints
and that's what we provided. So that also helped in
(25:31):
the reach for them to know that there is a
device out there which is very cost effective, which won't
cost thousands and lacks of rupees for them, just a
dollar which is a meal a day, right, So that
amount of money to protect their crops that was huge
for them. So that also helped us make them acquainted
with the technology and the benefits of it.
Speaker 1 (25:54):
At the cost of one dollar to use kishan No.
The America and Rishikish have made these resources accessible to
those who need it most, but being cost effective is
only half the battle. They had to work hand in
hand with the farmers to teach them how the technology worked.
But this technology had a more profound impact in identifying
(26:16):
the source of the crop loss. It also led to
revelations about the dangerous fertilizers and pesticides they were using.
How have you found the process of having the farmers
actually take some action based on the results that you
give them.
Speaker 5 (26:32):
Initially, like they didn't understand what exactly we were trying
to do.
Speaker 4 (26:36):
They just were, Okay, there's nothing harm in it, but
there's nothing good in it. So that's how it was.
So we actually startle if some visual based learning. Each
weekends we try to un make them understand what exactly
we were doing in just some graphics, cartoon type animations,
just to understand what exactly we are trying to do,
so that they're also getting literate about Okay, this is
(26:57):
a technology that they are paying for the cost of
for one acre of land in crop area was just
costing them around seventy troopees. That's around one dollar near
to one dollar, and it was a monthly based service,
so they were giving for each acre seventy troopees.
Speaker 5 (27:11):
Each farmer would have been paying us.
Speaker 4 (27:13):
The cost was just to handle out the server that
we were trying to maintain, and these informations that we
are trying to literate them with.
Speaker 5 (27:21):
They understood at least some parts of the technology.
Speaker 4 (27:24):
They understood how exactly the pest and plant disease affect
the crop, and what kind of pesticides, what kind of
fertilizers are actually affecting both the crops and both.
Speaker 5 (27:35):
As humans when we consume that product.
Speaker 4 (27:38):
So they also started to understand and they started to
stop using those pest sets and fertilizers for a particular
duration of time because in India, in particular crops, they
farmers just used to spray pesticides and fertilizers even if
they have not been attacked by any pests. This is
used to spray it before any pest infestation, just to
understand that it should be protected. But actually it's was
(28:00):
hampings as human beings because even if there is no
pest attack, we were actually consuming that pesticides and fertilizers.
Speaker 2 (28:08):
It matters on how we present the data to farmers,
and this also ties into the whole digital literacy programs
that we wanted to run. And as the Religash mentioned,
we were trying to present the data to them in
a way that they could understand as an individual. Anne
I impact enthusiast. I believe that having that AI accessible
in regional languages is very important and that is something
(28:31):
that we try to incorporate as well. And even as
Retigish mentioned, like pesticides, when used unnecessarily, they do drive
the costs also, so the farmers, if you don't talk money,
they do understand that, right, So you can see, you know,
like all the pesticides that you have been using, you
don't have to use those much. You can just use
(28:52):
on the base of the data that we're giving you,
and that too in a very accessible form.
Speaker 1 (28:56):
And Sweeta we talked a little bit about previously around
regulations and how Intel can assist the adoption of these
sorts of technologies. I mean, we heard from Risha, Kisha
and Erica that they had to sort of contact the
local administration bureaus to get permission. Maybe you could talk
a little bit about the way Intel can actually help
(29:18):
that process to get the technology down locally.
Speaker 3 (29:23):
So all countries governments, both at the central government level
and at the local government level today are developing strategies
on how do you really take emerging technology to the
last mile or to the grassroot level. Nindia specifically has
a very rapus Daia strategy on how do you really
develop a sustainable, inclusive, positive impact on citizens by improving
(29:47):
public awareness, by helping them access public services which would
allow technology to become a part of their regular routine.
Speaker 5 (29:56):
The way they work, the way they.
Speaker 3 (29:58):
Function, such as what Niharika and Nishikisha developed can be
driven in a larger way, can be scaled with the
help of the local state government and we're already working
with multiple state governments to ensure that they create platforms
where these can be taken further. The idea or the
objective of our collaboration with the government is how do
(30:19):
we really bring AI everywhere in an extremely inclusive and
responsible manner. But a large obstacle which I see is
the availability of infrastructure right because for the adoption of technology,
we have to make sure that precision farming requires investments
in digital infrastructure at scale and now there are multiple
(30:40):
schemes and initiators which coment to in India is doing.
They're trying their best to improve the living standards of
Indian farmers, trying to support them in smart farming practices.
But apart from this, there are tax benefits, there are
financial grants, etc. Which can help accelerate the cost of
technology adoption.
Speaker 1 (30:58):
In terms of AI, and it's becoming obviously more popular
across multiple industries. What's the number one thing you'd like
to try and solve using AI technology in the in farming.
I'll start with the Ahurica.
Speaker 5 (31:12):
Thank you for the question.
Speaker 2 (31:14):
So it's a wonderful question and I could think of
a million things that I could solve, and I'm pretty
sure the farmers would also agree with me. But one
of the things that I believe would be a very
huge issue that AI could potentially solve is protecting farmers
and their farms from climate change. Now, this is a
(31:35):
huge issue that's cropping. Our global climatic changes are worsening
every year. There's droughts everywhere, there's floods in some places,
So things like that farmers should be protected from natural
calamities disasters like that that could potentially just endanger their
livelihoods and destroy their economic and social levels, and that
(31:56):
is something that we should look into as AI enthusiast
on how to protect far from that, and that I
believe would be one way that AI could totally revolutionize
the agricultural industry.
Speaker 1 (32:07):
Excellent, Rishi, Kishi, You've had extra time to think, so yeah.
Speaker 4 (32:11):
So basically the area that I'm also currently working on,
that's the genomics selection of particular varieties in crop farms,
and that's one area that AI can be used to
analyze vast genomic data to identify genes associated with desirable
crop traits that can adapt to the climate change. Because
as you're proceeding, like we all know like where exactly
(32:33):
we are proceeding on, so the only way is to
adapt to the upcoming situations and to prevent it. So
I'm working on the adaption side of the climate change
in particularly farming. So we are trying to understand how
these AI tools and AI can be used. Machine learning
algorithms can be used to understand this various genomic data
and create new genomes that can actually accelerate breeding programs,
(32:54):
resulting in crops that are more disease resistant, nutritious.
Speaker 5 (32:58):
And adaptable to changing emitic conditions.
Speaker 4 (33:01):
So that's one area that can be a very huge
factor to revolutionize the farming sector.
Speaker 1 (33:08):
And Shwita, what's the number one area of AI technology
you'd like to see.
Speaker 3 (33:13):
Actually focus on most sustainable and economical farming which as
a result provides or becomes climate smart farming. So that
is where adoption of smart farming practices right, which would
really help grow India and the farmer and the community
to which they belong.
Speaker 1 (33:32):
Excellent, excellent. I would like to thank my guests Rishi
kish Ahmit Nayak, Swita Karuna and Niharika Haridas for joining
me on this episode of Technically Speaking and Intel podcast.
This was a very enjoyable discussion for me as I
love talking with actual innovators, engineers and developers with fruits
on the ground deploying technology and making a difference. What
(33:55):
amakes me about the efforts was the use of the
Intel Open Vino platform and it seemingly casual use of it.
It was only a few years ago that running machine
learning in AR models was a massive undertaking. The kishan
No initiative that Ushikish and Erica have developed is a
testament to the ability for new AI tools like Intel
open Vino to speed up the development and deployment of
(34:16):
industry changing technology. It was also important to understand the
larger governmental impact on AI development. We heard from our
guests the importance of ensuring that we strive to reduce
any barriers to innovators from exploring, experimenting, and succeeding in
their AI efforts democratization of technology. By continually striving to
reduce the cost of AI deployments, two tools like Intel
(34:37):
open Vino will be a boomed not only to the
remote villages of India, but also in the tallest skyscrapers
of New York. Please join us on Tuesday, October thirty
first for the next episode of technically Speaking, an Intel podcast.
(35:01):
Technically Speaking was produced by Ruby Studios from iHeartRadio in
partnership with Intel, and hosted by me Graham Class. Our
executive producer is Molly Sosher, our EP of Post Production
is James Foster, and our supervising producer is Nikair Swinton.
This episode was edited by Cierra Spreen and written and
produced by Tyree Rush
When a lot of us think of farming, it reminds
us of simpler times, and perhaps it feels like one
of the remaining industries exempt from the influences of the
modern tech world. But imagine a world where the success
of your family's farm crop yield is access to AI tools.
There's so much labor and effort that goes into maintaining
a farm, especially when farmers have to anticipate unpredictable weather
(00:25):
patterns and unprecedented seasons brought on by climate change. Plants,
like humans, are living things, with millions of tiny organisms
both attacking and assisting their life cycle. Some threats to
crop life are smaller than the human eye can see,
and when not addressed, the results can be disastrous to
local economies. But what if AI could solve the problem.
(00:46):
Giving eyes and access to where farmers cannot reach. AI
protects crops and the economy from the threat of microbial pests,
resulting in a more prosperous tomorrow. Hey there, I'm gram
Class and this is technically speaking an Intel podcast. The
show is dedicated to highlighting the way technology is revolutionizing
(01:08):
the way we live, work, and move. In every episode,
we'll connect with innovators in areas like artificial intelligence to
better understand the human centered technology they've developed. There has
always been a disconnect between nature and technology. However, today
there's a lot of science and technology at the core
of modern farming, and we're not talking about GMOs. One
(01:29):
of the biggest issues in agriculture is environmental threats. Farmers
struggle with protecting crops from diseases and pests without using
tools that could adversely affect consumers. AI has been instrumental
in helping farmers detect pests before infestations occur and result
in huge crop loss. But before we get into exactly
how it all works, I want to introduce our guests.
(01:53):
In twenty seventeen, Rishi kish amitt Nayak's family farm in
India so ninety percent crop loss due to pest in infestation.
After partnering with a fellow student, Niharika Haridas, the two Megatronics,
Robotics and automation engineering students found a way to use
AI to develop a method that could detect crop pests
(02:13):
through thermal imaging. This system, called kishan No, has been
proven effective and very affordable to local farmers. Rishikish America
thanks for being here.
Speaker 2 (02:23):
Such a pleasure to be here, Graham, thank you for
the invitation.
Speaker 1 (02:26):
We're also joined by Shwita Karuna, intel's Director of Government
Partnerships and Initiatives for Japan and the Asia Pacific. Sharita
has over twenty three years of experience creating trusted government
relationships and fostering government programs that encouraged the implementation of
modern science into the workforce PLUSH. She was instrumental in
(02:46):
helping kishan No grow as a farming tactic across the region. Welcome, Shwrita.
Speaker 3 (02:51):
Thank you, Graham, such a pleasure being here.
Speaker 1 (02:56):
So let's start at the beginning a very interesting story
around rishikish Can you tell a little bit about the
problem that your family and other farmers experienced back in
twenty seventeen.
Speaker 4 (03:07):
In India particularly, it's an agricultural country, so more than
seventy percent of the people do agriculture as their own
major occupation. In twenty seventeen, my father's grandfather was completely
invested into agricultural farming, and during that time, in Orisa particularly,
there was a plant best attack that couldn't be identified
for a longer period of time, and that resulted in
(03:28):
a lot of crop losses and hectares of land was
just lost because of an unidentified pist. Personally, we saw
a lot of farmer suicides in our own village, and
that was the major reason when I thought, Okay, I
do have a background of engineering, I do have a
background of robotics, so why not to create something for
our own farmers. And being part of that family where
(03:50):
we do farming in our parental site, I was just
touched with that fact that I need to do something
for the farmers.
Speaker 1 (03:58):
In Rishikishi's village alone, there were four farmers who took
their lives as a result of the devastated crop, and
his family saw a ninety percent crop loss that year.
The infestation was so devastating to their livelihood his family
considered leaving farming all together. And to make matters worse,
the problem was difficult to identify and trace. Before we
(04:23):
get into the actual details of how you solved it
in Arika, how did you get involved in the project.
Speaker 2 (04:30):
I decided to pursue mecatronics and automation at Viatchene out
of a sheer passion for robotics as a twelfth grader.
So I came across the work that many companies like
Boston Dynamics were doing at that point and exactly right
the spot pro vote of course, and I was just
enthralled with potential that it helped, Like it was like,
(04:50):
oh my, what this could change humanity?
Speaker 5 (04:52):
And I was like, I need.
Speaker 2 (04:54):
To do something in this space. I wanted to help
people with this new technology. And that's how I went
to Aitchen and that's where I'm Metri Shikish and we
started talking and we were talking about this project and
I was like, you know, that's that's amazing that we'll
let me contribute to it as well, and that's how
we started collaborating on the project and then we participated
(05:14):
in the Inaugril Intelliet Global Impact Festival and the rest
is history. We had a wonderful time and you know,
the support that we have gotten from Intel for it
as well has been phenomenal and that's the reason that
Kishano is at the place where it is right now.
Speaker 1 (05:29):
Excellent. So now as the I guess the sixty four
thousand dollars question is how does the kishan No work.
Speaker 4 (05:37):
Kishano basically taps into saturate based thermal imagery. These images
can detect temperature variations and crops which often indicate Streuss
disease or pestal activity. For instance, areas affected by certain
pest or microbol infestations may exhibit different thermal patterns compared
to healthy areas. We collect images from Sentinel two and
lands At eight satellites. Those satellite images are then sys
(06:00):
to get index mapping out likes, for example, vegetative indexes
and moisture indexes through a software called QGIS, so it
basically gives us the values for those vegetative indexes and
moisture indexes, and these gathered thermal imageries processed using AA algorithms,
where we've processed the images first into the open Veno
platform and we get a d blood image for better
(06:23):
accuracy of training of the models. Then these algorithms are
trained to recognize patterns or animalies that correspond to microbilan
pest outbreaks. Over time, has more data is collected and analyzed,
the AA model becomes more accurate and efficient in its
prediction and leveraging the power of machine learning. Once a
potential threat is identified in the system, the systems can
(06:43):
send alerts or recommendations to the farmers in the local
administrative levels, where we also design the physical device apart
from the AA algorithm to get a confirmatory test that
there is a pest or plant disease outbreak. This actually
includes information about the type of threat, it's severe, and
recommendation algorithms.
Speaker 5 (07:02):
This proactive approach.
Speaker 4 (07:03):
Helps farmers to take actions before the problem becomes widespread
and saving both time and resources.
Speaker 1 (07:10):
I'd like to talk about Intel open Veno a little
bit so quickly, just to inform our audience. INTE open
Vino is a cross platform toolkit developed by Intel that
deploys deep learning models on visual data sets, helping computers
better recognize and process images to inform decision making. But
I'm curious as someone who's just as interested in what
(07:30):
didn't work as opposed to what ultimately does. Why did
you decide to use Intel open Veno. Were there are
other methods you tried first?
Speaker 2 (07:38):
So we did try a lot of techniques, and we
found that open Veno worked perfectly with our project, especially
with the goal that we were trying to achieve. So
we saw that the hardware requirements as well as the
software requirements did completely match. Also, we had mentorship from
Intel and we were able to properly and in a
better way adapt to those systems to our project, and
(08:01):
that's the reason which it was open.
Speaker 4 (08:03):
We know, we actually tried to degler images through some
deep learning algorithms, but those algorithms was actually not satisfying
the accuracy that we actually wanted, so open Veno just
suited out the case perfectly.
Speaker 1 (08:17):
One thing I'm interested in is the pests that were
being detected. Am I right in saying that it had
a unique therm signature?
Speaker 5 (08:24):
Yeah?
Speaker 1 (08:24):
And how did you discover that?
Speaker 4 (08:27):
In twenty seventeen, Once we identified the problem, we actually
tried to create a physical device through a thermal camera
set up and microprocesses. We were rotating that device among
the periphery of the crop fields to understand what exactly
the thermal traces are in the leaf of the crop plants.
And once we understood what are the thermal signatures for
(08:47):
different crop plants, we understood there is a concept that
whenever there is a pathogen or a plant disease, there
is a certain increase in the leaf temperature. And if
we identify that leaf temperature increases in the particular or
in a particular duration of time, we can actually significantly
say that there is a best attack or a plant
disease in the crop area. Once we had the theory,
(09:10):
we tried to incorporate that similar formula in the vegetative
index of the satellite setup. So in twenty nineteen we
had the physical setup, we tried the same literature to
understand it to the satellites.
Speaker 1 (09:22):
Hearing Rishikish and Aharika elaborate on how they design their
imaging tool reminded me of my own experience attempting to
develop systems to work remotely in the jungles of Africa.
It's not an easy feat though, as there's no real
infrastructure for these sorts of products, especially when they are
limited by internet access and availability in the area. Hearing
(09:44):
how much progress these two had made with their program,
maybe wonder about the challenges that went into making this
tool available in the rural farmlands of India.
Speaker 2 (09:57):
There has always been a digital divide in India, as
we can see, but now it's been narrowing and that's
a very good news for all of us, and that
infrastructure is also becoming better. There's also research that India
has one of the cheapest internet out there in the world,
so I mean, it's being adapted and we are glad
that it is. But when we were working on it,
(10:17):
we did face a lot of infrastructure issues regarding internet
services as well and internet connectivity exactly, and sort of
having that satellite imagery. Gaining access to the satellite imagery
was very difficult for us because that area wasn't mapped.
Remote areas aren't usually mapped with that much precision as
that of let's say, an urban area, so we did
(10:38):
have some issues with that, but then we did try
our best to solve those and gain satellite images from
the areas that we neated.
Speaker 4 (10:46):
Farmers in the villages particularly, they were quite a bit
skeptical to try this out, and the farms because in
India particularly didn't back that time, we didn't have that
much of agritechnology tools or products, and going as a
youngster something around in class ninth or tenth and trying
out as some different new projects or new census in
(11:07):
the field, they were quite a bit skeptical. So managing
that side of that, Okay, we are doing something good,
we are doing something better for your own crops, we
are doing something for the best of the society. Convincing
them was one of the very huge challenge over there
in India.
Speaker 1 (11:23):
What kind of data or training processes were involved in
training the model to recognize microbiopests in the crops.
Speaker 5 (11:31):
Initially it was only deep learning algorithms.
Speaker 4 (11:33):
Further on, when we had a lot of thermal praise
data and we had did the d blood images, we
were just focused on the CNN models to train the data.
And it hadn't given a good accuracy of for around
ninety points something percentage, so it was a pretty good
accurate to start with for a particular set of crops.
Speaker 1 (11:51):
You said, CNN, could you just define what that is please?
Speaker 5 (11:54):
Conventional neural network.
Speaker 1 (11:56):
Okay, And that's just another AI technique to for learning.
Speaker 5 (12:01):
Yeah, yes, a machine learning okay, okay.
Speaker 1 (12:04):
And you just mentioned about the accuracy that you achieved.
Would you say that's typical for the Intel Open Veno
platform to get that sort of result.
Speaker 4 (12:14):
The accuracy is for the total accuracy of the model
for a particular set of crops, for example, tomatoes and wheat.
For those two crops we had an accuracy fround ninety
point two eight percentage, and for other crops it's still
in the process of getting more accurate and all. So
for these two crops, overly, it was the accuracy that
we measured out and.
Speaker 1 (12:33):
In terms of the Intel Open Veno technology, can you
think of anything any other farming use cases beyond pest
management and crop protection.
Speaker 5 (12:42):
Currently, we were trying to work on crop genome analysis
where we were actually trying to understand because of the
climate change to the new variants of crops are needed
to adapt to the new climatic conditions. So we were
trying to understand how exactly we can use machine learning
algorithms to create new genomes in the crops the microbiology
side of it.
Speaker 4 (13:02):
So yeah, that's one area that I was completely focused
on in this past recent days.
Speaker 2 (13:08):
I would like to add on to that, and as
Education mentioned, convolutional neural network model that we used, it
was at that point not something that was used by
the AI community, but then we now see a lot
of use cases for that and that's something that we
are very glad about. And also some of the use
cases that I have at least found as an AI
enthusiast that models like these could have is in real
(13:32):
time data, especially as the climatic change has become a
huge issue. It is something that can help a lot
of farmers with when there is excessive rains or when
there is no rain at all, to predict these through
AIML technologies. And I believe that the limit is boundless
when it comes to AI technologies. Right we are seeing
a start of a new era of AI, and I
(13:54):
am very glad to see how I was being used
by lots of companies, and we also hope to go
contribute to that, and I hope for a very bright future.
Speaker 1 (14:06):
AI has been the focus of a lot of discourse
over the last couple of decades. While many of us
experience it as virtual assistance in our phones and computers,
AI has been giving us listening, watching, and reading recommendations
for years and we continue to see it evolve and
even create content like images and written stories. But that's
(14:28):
all just the beginning. AI has so much potential to
positively impact the way we work and live. It can
be used to detect new variants and threats in agriculture
brought on by climate change conditions. The Intel Open Vino
technology played an essential role in this, providing higher accuracy
for detection. I'd just like to switch a little bit
(14:51):
to the agribusiness side of things. And maybe I can
get Shuita to comment on this in terms of the
Intel Open Vino and its app cation here for pest detection.
Do you see it complementing other pest control methods in
agriculture and does it have the potential to replace pesticides
and insecticides and farming replace.
Speaker 5 (15:13):
Is a little on the harsher terms.
Speaker 3 (15:16):
What I would actually look at it is AI and
agricultures really helping farmers make data driven decisions, optimize crop
yields conserved resources like water and energy. The challenge here
is not just the solution part of it is also
kind of encouraging next generation technologists student innovators to come together,
look at the local problems like what Neharikan risikation have done,
(15:40):
and then create a solution using all the skills they've
learned as part of their formal education as well as
as part of being a part of Intel programs the
Interdigital Rediness Program portfolio, come together and democratize AI skills
in a way which gets a common person a farmer,
to understand trust and emergingology like artificial intelligence and hopefully
(16:02):
become comfortable in applying it to solve the daily problems
they would be facing as part of their community.
Speaker 1 (16:09):
I love that term democratization of technology, and I think
that's ultimately what technology does is get it more accessible
and cheaper to get it to the far regions of
the world. I'd just like to expand a little bit more,
maybe if you could explain some of the programs that
are available through inter Open VENO to help farmers or
(16:29):
businesses with limited resources to get access to this sort
of technology and expertise.
Speaker 3 (16:35):
I'll just take a step back here, right because we
keep talking about increasing digitization, which today a lot of
governments and countries are going towards. But what it really
means is when we focus on increased digitization, we also
need to invest more in building the digital readiness of people,
specifically in terms of emerging in critical technologies like AI
(16:55):
or what you spoke about, like the usage of open Wino.
How do we get person to understand how they can
utilize the technology like open we know to be able
to solve their local problem and create indigender solutions. So
all this kind of comes together through a whole program
portfolio which we have which is called the Intel Digital
Readiness Programs, which is driven in collaboration with government, academia,
(17:19):
civil society, and the industry and focuses around building shared value,
shared responsibility so that we can really demystify democratize these
superpowers which we keep talking about, like artificial intelligence for
a very broader and a diverse audience for young budding
technologists like Neiharika Ushikish but also for those who are
(17:42):
going to be consuming the technology at the other end
of the spectrum. The programs are a lot, they're many.
They range from you know, programs like AI for Citizens,
which talks about getting a citizen to understand how to
navigate his or her life in an AI driven world.
AI for Youth, which really allows us to empower youth
with not just the technical skills associated with AI, but
(18:05):
also the social skills in a very inclusive manner. And
then we have AI for Future Workforce, which is for
vocational community college students, engineering students, which really helps them
to understand how to be prepare themselves for becoming a
part of the future workforce. So a huge spectrum, lots
of programs, but the one which is very special to
all three of us in this case, and I'm sure
(18:27):
Education Aherka would agree with that is our EI Global
Impact Festival, because this is where we work with all
these student innovators. We get them together and we get
them to celebrate their AI innovations with a huge passage
which does not just allow them to showcase what they've built,
but also helps them hone their skills by getting mentored
(18:48):
by Intel technologists because at the end of the day,
these young students are the next generation technologists, so we
want to make sure we work for them to support
and build an AI ready generation.
Speaker 2 (19:00):
Platforms like these have been really instrumental and I have
seen the impact on ground that they make in supporting technologists,
young technologists like us, and we have always been very
grateful for the opportunities and mentorship as well that Intel
has provided. And that's something that we wish that every
budding technologist in India and all over the globe can
(19:22):
at least experience, because mentorship and guidance is an important
pillar of one's journey and having someone who can teach
you more about AI, how to use AI, and how
to benefit from AI, especially with the immense potential it
has that is life changing.
Speaker 1 (19:42):
You're listening to technically speaking, an Intel podcast will be
right back. Welcome back to technically speaking an Intel podcast
(20:03):
shweeta last episode of this podcast, we talked with Reachhuvu,
one of your colleagues, about the ethics and responsibility of AIM,
wondering if we could get your thoughts on how you're
working with governments and industry leaders around AI and trying
to help them navigate some of the ethics and responsibilities
(20:24):
around AI development.
Speaker 3 (20:26):
That's a very interesting question for us, right because when
we speak about digital reddiness or how do we build
digital readiness, we look at three pillars. Largely, one is,
of course learning the skills of emerging technologies like AI,
but more importantly, getting to understand and trust those skills,
So getting to understand not just what the advantages are,
(20:47):
but also what the pitfalls are. Getting to understand which
situation should we apply the emerging technology in and which
ones we should abstain from. So our programs, in fact,
inculcate a lot of discussions around these there is, which
range from the ethics piece of it, which range from
how how do we make it more inclusive, how do
we make it more diverse? And so much so that
(21:10):
if you kind of package it all together, it comes
under the larger umbrella of responsible AI. So how do
we really encourage not just youth, but every citizen, which
includes the governments who we collaborate with and partner with
to understand what is the responsible use of these superpowers
like AI to gain broader socioeconomic benefits for everybody.
Speaker 2 (21:32):
As a youth igffellow. That is exactly what I focus
on Internet governance right and how AI governance works and
how AI can be regulated. But then what about AI innovation?
It shouldn't be regulated or stifled due to laws coming
into place that can have that effect where people continuate
and they can't contribute to new technologies, so that there's
(21:55):
a delicate balance between them, and that is what I
also do look into. And the whole area of how
becoming emerging technology is like even robotics which has a
huge artificient intelligence ethics background out there, So how do
we harness this without harming humanity? And that is something
that I believe all stakeholders, including the youth of our
(22:16):
country or the globe, should be focusing on because there
also tends to be the whole bias of youth not
being given a voice when it comes to these emerging technologies.
But I believe if they do understand what it is
about and what potential risks it has and what potential
benefits it has, that gives them the knowledge to use
it responsibly and ethically.
Speaker 1 (22:39):
Using AI can be as complicated as Niharika has pointed out,
but the tool she and Wishikish have been able to
create from this place of innovation and AI have changed
the world for the better and they have the results
to prove it. In terms of the Kisheno technology that
you have developed, do you have any stats on the
(23:02):
crop that has been saved or the reduction in crop loss?
Speaker 4 (23:06):
In twenty to twenty, we actually piloted this around in
eight districts in Orissa and more than around seventy two villages.
We actually serve it upon and piloted upon and for
one season we tried it particularly on wheats and tomatoes.
Once we had data that we could actually predict that
there is a pest attack or plant this is coming up,
(23:27):
we use that data to try to save those fifty villages.
We used pesticides and fertilizers just before whenever the pest
and pest attack could have happened, So it actually saved
around those fifty villages.
Speaker 1 (23:40):
I'm really interested in how the technology actually is deployed
and distributed to as many villages as possible. To me,
the innovation is part of that as well. How do
you deploy it, how do you scale it? And you
said you went to seventy two villages, how did you
get to all of them and provide this service and
this knowledge.
Speaker 5 (24:00):
In the local districts.
Speaker 4 (24:01):
We contacted the administrations and with the recognitions we had
with until it was really easy to contact the administrations.
So once we had contacted the administration the local villagers, they
were actually understood, Okay, there is someone who is coming
to do something in their villages and it won't harm them,
So they were at least a relaxed that nothing is
going to be happening.
Speaker 5 (24:20):
And also they actually co operated out.
Speaker 4 (24:22):
So we had to draw the plots, We had to
map it on the satellite software that we had and
it would actually give us a satellite based crop image.
And for each crop images, we just needed to market
around the perimeters of that particular individual farmer and the
work is done. We just needed to understand how what
area that particular farmer has, what is the crop type?
(24:44):
When did so what is the raining patterns and what
is the soil type. With these certain parameters understood, the
farmer had to do nothing. We were sitting on a
room played server and we were training these images and
it was again the process kept on going. We had
the results each week, we just to share them. Okay,
this is the condition, this is what your crop health is,
and your crop is safe and if not, we will
(25:07):
at least give them some predictions.
Speaker 2 (25:09):
One of the other plus points or advantages of our
innovation was how cost effective it was. So now this
is a huge issue when it comes to India that
technologies are out there, but they can be very expensive
and that's not reachable to a conventional Indian farmer. They
need solutions that are cost effective because of budget constraints
and that's what we provided. So that also helped in
(25:31):
the reach for them to know that there is a
device out there which is very cost effective, which won't
cost thousands and lacks of rupees for them, just a
dollar which is a meal a day, right, So that
amount of money to protect their crops that was huge
for them. So that also helped us make them acquainted
with the technology and the benefits of it.
Speaker 1 (25:54):
At the cost of one dollar to use kishan No.
The America and Rishikish have made these resources accessible to
those who need it most, but being cost effective is
only half the battle. They had to work hand in
hand with the farmers to teach them how the technology worked.
But this technology had a more profound impact in identifying
(26:16):
the source of the crop loss. It also led to
revelations about the dangerous fertilizers and pesticides they were using.
How have you found the process of having the farmers
actually take some action based on the results that you
give them.
Speaker 5 (26:32):
Initially, like they didn't understand what exactly we were trying
to do.
Speaker 4 (26:36):
They just were, Okay, there's nothing harm in it, but
there's nothing good in it. So that's how it was.
So we actually startle if some visual based learning. Each
weekends we try to un make them understand what exactly
we were doing in just some graphics, cartoon type animations,
just to understand what exactly we are trying to do,
so that they're also getting literate about Okay, this is
(26:57):
a technology that they are paying for the cost of
for one acre of land in crop area was just
costing them around seventy troopees. That's around one dollar near
to one dollar, and it was a monthly based service,
so they were giving for each acre seventy troopees.
Speaker 5 (27:11):
Each farmer would have been paying us.
Speaker 4 (27:13):
The cost was just to handle out the server that
we were trying to maintain, and these informations that we
are trying to literate them with.
Speaker 5 (27:21):
They understood at least some parts of the technology.
Speaker 4 (27:24):
They understood how exactly the pest and plant disease affect
the crop, and what kind of pesticides, what kind of
fertilizers are actually affecting both the crops and both.
Speaker 5 (27:35):
As humans when we consume that product.
Speaker 4 (27:38):
So they also started to understand and they started to
stop using those pest sets and fertilizers for a particular
duration of time because in India, in particular crops, they
farmers just used to spray pesticides and fertilizers even if
they have not been attacked by any pests. This is
used to spray it before any pest infestation, just to
understand that it should be protected. But actually it's was
(28:00):
hampings as human beings because even if there is no
pest attack, we were actually consuming that pesticides and fertilizers.
Speaker 2 (28:08):
It matters on how we present the data to farmers,
and this also ties into the whole digital literacy programs
that we wanted to run. And as the Religash mentioned,
we were trying to present the data to them in
a way that they could understand as an individual. Anne
I impact enthusiast. I believe that having that AI accessible
in regional languages is very important and that is something
(28:31):
that we try to incorporate as well. And even as
Retigish mentioned, like pesticides, when used unnecessarily, they do drive
the costs also, so the farmers, if you don't talk money,
they do understand that, right, So you can see, you know,
like all the pesticides that you have been using, you
don't have to use those much. You can just use
(28:52):
on the base of the data that we're giving you,
and that too in a very accessible form.
Speaker 1 (28:56):
And Sweeta we talked a little bit about previously around
regulations and how Intel can assist the adoption of these
sorts of technologies. I mean, we heard from Risha, Kisha
and Erica that they had to sort of contact the
local administration bureaus to get permission. Maybe you could talk
a little bit about the way Intel can actually help
(29:18):
that process to get the technology down locally.
Speaker 3 (29:23):
So all countries governments, both at the central government level
and at the local government level today are developing strategies
on how do you really take emerging technology to the
last mile or to the grassroot level. Nindia specifically has
a very rapus Daia strategy on how do you really
develop a sustainable, inclusive, positive impact on citizens by improving
(29:47):
public awareness, by helping them access public services which would
allow technology to become a part of their regular routine.
Speaker 5 (29:56):
The way they work, the way they.
Speaker 3 (29:58):
Function, such as what Niharika and Nishikisha developed can be
driven in a larger way, can be scaled with the
help of the local state government and we're already working
with multiple state governments to ensure that they create platforms
where these can be taken further. The idea or the
objective of our collaboration with the government is how do
(30:19):
we really bring AI everywhere in an extremely inclusive and
responsible manner. But a large obstacle which I see is
the availability of infrastructure right because for the adoption of technology,
we have to make sure that precision farming requires investments
in digital infrastructure at scale and now there are multiple
(30:40):
schemes and initiators which coment to in India is doing.
They're trying their best to improve the living standards of
Indian farmers, trying to support them in smart farming practices.
But apart from this, there are tax benefits, there are
financial grants, etc. Which can help accelerate the cost of
technology adoption.
Speaker 1 (30:58):
In terms of AI, and it's becoming obviously more popular
across multiple industries. What's the number one thing you'd like
to try and solve using AI technology in the in farming.
I'll start with the Ahurica.
Speaker 5 (31:12):
Thank you for the question.
Speaker 2 (31:14):
So it's a wonderful question and I could think of
a million things that I could solve, and I'm pretty
sure the farmers would also agree with me. But one
of the things that I believe would be a very
huge issue that AI could potentially solve is protecting farmers
and their farms from climate change. Now, this is a
(31:35):
huge issue that's cropping. Our global climatic changes are worsening
every year. There's droughts everywhere, there's floods in some places,
So things like that farmers should be protected from natural
calamities disasters like that that could potentially just endanger their
livelihoods and destroy their economic and social levels, and that
(31:56):
is something that we should look into as AI enthusiast
on how to protect far from that, and that I
believe would be one way that AI could totally revolutionize
the agricultural industry.
Speaker 1 (32:07):
Excellent, Rishi, Kishi, You've had extra time to think, so yeah.
Speaker 4 (32:11):
So basically the area that I'm also currently working on,
that's the genomics selection of particular varieties in crop farms,
and that's one area that AI can be used to
analyze vast genomic data to identify genes associated with desirable
crop traits that can adapt to the climate change. Because
as you're proceeding, like we all know like where exactly
(32:33):
we are proceeding on, so the only way is to
adapt to the upcoming situations and to prevent it. So
I'm working on the adaption side of the climate change
in particularly farming. So we are trying to understand how
these AI tools and AI can be used. Machine learning
algorithms can be used to understand this various genomic data
and create new genomes that can actually accelerate breeding programs,
(32:54):
resulting in crops that are more disease resistant, nutritious.
Speaker 5 (32:58):
And adaptable to changing emitic conditions.
Speaker 4 (33:01):
So that's one area that can be a very huge
factor to revolutionize the farming sector.
Speaker 1 (33:08):
And Shwita, what's the number one area of AI technology
you'd like to see.
Speaker 3 (33:13):
Actually focus on most sustainable and economical farming which as
a result provides or becomes climate smart farming. So that
is where adoption of smart farming practices right, which would
really help grow India and the farmer and the community
to which they belong.
Speaker 1 (33:32):
Excellent, excellent. I would like to thank my guests Rishi
kish Ahmit Nayak, Swita Karuna and Niharika Haridas for joining
me on this episode of Technically Speaking and Intel podcast.
This was a very enjoyable discussion for me as I
love talking with actual innovators, engineers and developers with fruits
on the ground deploying technology and making a difference. What
(33:55):
amakes me about the efforts was the use of the
Intel Open Vino platform and it seemingly casual use of it.
It was only a few years ago that running machine
learning in AR models was a massive undertaking. The kishan
No initiative that Ushikish and Erica have developed is a
testament to the ability for new AI tools like Intel
open Vino to speed up the development and deployment of
(34:16):
industry changing technology. It was also important to understand the
larger governmental impact on AI development. We heard from our
guests the importance of ensuring that we strive to reduce
any barriers to innovators from exploring, experimenting, and succeeding in
their AI efforts democratization of technology. By continually striving to
reduce the cost of AI deployments, two tools like Intel
(34:37):
open Vino will be a boomed not only to the
remote villages of India, but also in the tallest skyscrapers
of New York. Please join us on Tuesday, October thirty
first for the next episode of technically Speaking, an Intel podcast.
(35:01):
Technically Speaking was produced by Ruby Studios from iHeartRadio in
partnership with Intel, and hosted by me Graham Class. Our
executive producer is Molly Sosher, our EP of Post Production
is James Foster, and our supervising producer is Nikair Swinton.
This episode was edited by Cierra Spreen and written and
produced by Tyree Rush
Popular Podcasts
Dateline NBC
Current and classic episodes, featuring compelling true-crime mysteries, powerful documentaries and in-depth investigations.
The Nikki Glaser Podcast
Every week comedian and infamous roaster Nikki Glaser provides a fun, fast-paced, and brutally honest look into current pop-culture and her own personal life.
Stuff You Should Know
If you've ever wanted to know about champagne, satanism, the Stonewall Uprising, chaos theory, LSD, El Nino, true crime and Rosa Parks, then look no further. Josh and Chuck have you covered.