business

AI for plant diseases [#CEBiT inspiration]

5 Jun 2017

Przemyslaw Grzywa: Hi Pierre. We are at CEBiT – biggest technology fair in Europe. You have been awarded CEBiT Innovation Award, my congratulations. Please tell us what is PEAT and why it is so innovative.

Pierre Munzel: PEAT is a young start up from Hannover, Germany. We are making image recognition software for the automatic detection of plant damages for users of our app Plantix. We created a database with over 500,000 pictures of plant diseases and the database is growing everyday. Thanks to complicated algorithms and our knowledge about machine learning (the ability of computers to learn without being explicitly programmed) we are able to train the network to do the identification of plant diseases just by analyzing one picture. You take a picture in your garden and the app gives you a result within seconds, tells you what is the problem and how you can get rid of it, as simple as that.

How did you start the company?

We obviously started with an idea that came straight from our background at the university. Four of us participated in a university project in Brazil. We noticed that people are very interested in knowing what kind of what is wrong with their plants and how to help them.. They called a, at this time spreading plant diseases “morto subito” – “fast dead”. We noticed that it is difficult to find informations in the internet when you just know it is just called fast dead. So what are you going to see when you type “fast dead”? You will find pictures of car crashes and car accidents. Dead end.

Finally, we ended up with a machine learning thing because we were working with 3D models. What we got was, on one hand, an optical pattern that a plant expert can see with his eyes and we have self-learning algorithms so you are able to learn your patterns and with the use of AI – the magic came in. So we would like to transport knowledge of experts, things that they are able to find working in the field using their eyes (identification, giving information about treatment possibilities or what to do to handle a particular disease). We are transporting it into a mobile device, into a digital solution. The knowledge of an expert everywhere you go in the fields.

 I understand is it is a very niche market, right? Who is your customer and what is your business model?

I don’t think it is a niche market because agriculture is a global market. We are now living in 2017 but in around 2050 we will have about 10 billion people on this planet and they will need to be fed. On the other hand we have a loss of 15 to 30% of the annual yield wordwide and our mission is to help to avoid food scarcity on this planet. We see it on a global scale. On the other hand smaller farmers lack professional tools in developing countries. We want to transport this knowledge to people who are growing it for themselves, to feed themselves and their families. So if one day you have a loss because of plant diseases and a loss of the yield, then begin a big struggle.

We are focusing also on producers of plant protection services or suppliers of land machinery etc., who can also use our software, API, can integrate it into their own solutions to give our service of automated detection of plant diseases to their customers so they have benefit to order it. This app adapts to our customers, the app is free for private users and smaller farmers but we have also a B2B licensed model so providers of other apps or other solutions can integrate our solution into them.

Which of these is your focus right now? Which of these groups?

So the main focus now is on small farmers and on the private users of the app and we ajust launched Planitx in India. Plantix in Germany alone has more than 60,000 dwonloads at the moment, we have more than 100,000 sessions From the beginning, we wanted to do it on a global scale and we want to go to international very fast so we are starting out in India, working together with corporate partners over there. Therefore in 2017, we are focusing on small farmers because it’s like a symbiotic situation, we are helping the farmers to get rid of their problems and at the same time they are sharing the pictures of their infected plants with us and this helps to train our systems, to train our algorithms. We believe that if you want to feed the world, you need to include B2B large scale agriculture as well.

 Can you use the app on any plant? Is there any limitation? 

 

Theoretically, you can use it on every plant. You can use it on everything that has a visible pattern. But the challenge is to grow the right data base. Because you need hundreds and thousands of pictures of the same disease to teach the algorithm. This is our unique thing – we are the only one on this planet who has this big data base, simply because nobody did it before us. So, in case of a human recognition: somebody has to go to the field and tries to guess the proper disease using examples from a book. Our software needs a thousand of pictures of the same disease to become an expert.

So, what is the future for you?

The future for us – millions of people are using Plantix and share there information. When we have enough users of Plantix we can install an early warning system. Especially in rural areas. I can imagine a situation when a Plantix user takes a picture of the plant diseases, the pictures then come to us with GPS coordinates and inform us where is the problem and what is the problem. We plan to combine it with weather forecasting platform. We want to do that to be able to tell people “bad weather is coming, be prepared”. Definitely we would like to build the global biggest data base for plant diseases. And actually, we want to reach the point that every picture that have to do with a plant disease is shared on the PEAT service. This is one of the goal that we would like to achieve within the next years. We would like to define a new standard for image recognition for plant diseases.

 

PEAT team:

 

PEAT_Team_color

Related articles

Browse all