Presentation
-
Subject
Increase the incomes of small-scale food producers.
-
Technological module
Artificial Intelligence
A. Explanatory Research
-
Local Food Node
-
Project carriers
The local food nodes’ developers make the website usable and are therefore the main project carrier. This project would not exist without the help of the donors as donations are the only source of income.
-
Beneficiaries
The main beneficiaries of this project are small local food producers and consumers from Sweden as they can sell and buy the products. The secondary beneficiaries are the workers at the local delivery spot – node, and the environment. Having nodes creates the need for new jobs, and the environment benefits from the reduced CO2 emissions thanks to local deliveries.
-
Users
The typical users of Local Food Nodes are growers and customers from Sweden. The latest use the website to order food from the growers directly. The donors can also be considered as users since they use the website to send their donations.
-
Need
Local Food Nodes responds to the need of consumers to receive fresh local food as well as the need of growers to sell their products with a margin high enough to make a living. To make that happen the two parties need to be connected though a transparent process.
-
Principle
To connect suppliers and customers, Local Food Nodes creates local delivery spot called nodes, where the growers deliver the products the clients ordered. Anyone can create a node on the website! This solution is good for the growers’ income; delivering in only one location in a sector cuts off a good chunk of the delivery fees, and the food is directly delivered to the end user so there is no middleman! This allows the suppliers to increase their margin on every product. Additionally, Local Food Nodes is open source and displays its own transaction history on the website. The donation system is then completely transparent.
-
Main Technologies Involves
- User’s computer and smartphone
- Web coding languages: HTML/CSS, Javascript
- Open Source
- Electronic payment: Swish, Stripe
- GPS
-
Sources
- https://zaunders.net/LFN
- https://localfoodnodes.org/en
- https://stripe.com/docs
- https://developer.swish.nu/
- https://localfoodnodes.org/fr/team
- https://localfoodnodes.org/images/facebook-share.jpg
-
-
Cropswap
-
Project carriers
The Cropswap’s developers make the App and the website usable and are therefore the main project carrier. This project would not exist without the partnership between Cropswap and the farmers that offer their products on the App. The state of California granted 4M$ to Cropswap from the California Budget Act 2021, a funding without which the company could not have reached this level of importance in L.A.
-
Beneficiaries
The main beneficiaries of this project are small local food producers and consumers from around L.A as they can sell and buy the products on the App. The secondary beneficiaries are the workers who deliver the products at the door or in pick-up spot, and the environment. The delivery service creates the need for new jobs in the community, and the environment benefits from the reduced CO2 emissions thanks to local deliveries. Lastly, Cropswap encourages inhabitants of L.A to convert their outdoor spaces to gardens where they can grow a wide array of fruits and vegetables. This is positive for the biodiversity and contributes to make L.A greener.
-
Users
The main users of Cropswap are growers and customers from L.A, but the App is available worldwide. The customers use the App to order food from the growers directly.
-
Need
The project is filling the need of consumers for local and diverse food easily accessible. This goes along with the need of small food producers to increase their margins by selling directly to the customers with no middleman taking most of the profit away. Therefore, the combined need of suppliers and customers is to be connected.
-
Principle
Cropswap’s App enable farmers to register on their platform to list their products for the customers to buy. The subscription feature was introduced to make income more stable; it enables farmers able to plan ahead! The delivery process is completely transparent as you can follow your products on the App from the farm to your door. The are no middleman involved and the growers get 100% of the delivery fees. Cropswap still make money by taking 10% of the customers purchases and 8% of farmers transactions. However, farmers can set their prices as they wish. The firm also uses AI to help the growers streamline their operation. The AI is mostly a recommendation system based on preferences and seasons to the user.
-
Main Technologies Involves
- User’s computer and smartphone
- Web coding languages: HTML/CSS, Javascript, Swift
- iOS
- Android
- Electronic payment
- GPS
- Servers
- Database
-
Sources
- https://afrotech.com/crop-swap-la-receives-4-million
- https://www.prnewswire.com/news-releases/the-instacart-for-local-produce-farm-to-phone-app-cropswap-connects-communities-through-food-301035538.html
- https://cropswap.com/
- https://cdn.newswire.com/files/x/b2/0e/4cff8bad9f9c9de7eecfb5f782a6.jpg
-
-
Bristol Food Producers
-
Project carriers
Bristol Food Producers is a community benefit society. Its team and its members carry the project. Additionally, the Bristol city council sponsors the project by giving the company a grant.
-
Beneficiaries
The main beneficiaries are the farmers that receive help from Bristol Food Producers. The secondary beneficiary is the city of Bristol, which see its local food businesses thriving and its parcels of unused lands becoming green.
-
Users
The users are also the farmers of Bristol, they are the one paying the membership to become a member of the cooperative.
-
Need
There is a need to Improve fairness and efficiency for farmers in Bristol as they struggle to find land and to sell their products to markets at a good price.
-
Principle
Bristol Food Producers has four main points with which it helps farmers to get better income and better at their job. The company provides training to farmers and help them learn new skills to become more efficient. Secondly, the cooperative connects landowners and farmers as they usually struggle to acquire a small parcel of land by themselves. Thirdly, Bristol Food Producers connects farmers with customers directly, with restaurants, or with other retailers. Lastly, BFP develops solidarity between members by making them work together. Additionally, members subscribe to an online newsletter which goes hand to hand with the training. The memberships cost from 20£ to 50£ a year. For that price BFP helps small food producers develop a sustainable business.
-
Main Technologies Involves
- User’s computer and smartphone
- Web coding languages: HTML/CSS/JS
- Electronic payment Solution
- Database
-
Sources
- https://bristolfoodproducers.uk/our-work/network-solidarity/
- https://bristolfoodproducers.uk/wp-content/themes/docandtee-2017-new/assets/images/bfp-logo.png
-
-
Plantix
-
Project carriers
The main carrier of Plantix is the husband-wife duo Rob Strey and Simone Strey under their startup Progressive Environmental and Agricultural Technologies (PEAT). The company’s developers and data engineers are the main workforce of the project. The community that provides images to Plantix are also necessary to the project as the feed the database and help the AI to improve.
-
Beneficiaries
The main beneficiaries of the project are the small farmers from developing countries, with limited funds and limited land. More generally, every small farmer benefit from the project. Additionally, the local inhabitants living near the farmers have a better access to food thanks to Plantix as it helps prevent crop loss and thus increase the food supply.
-
Users
The users are mainly farmers with small parcels of land coming from developing countries. Secondly, bigger companies use Plantix premium services, “Plantix Analytics” and “Plantix Vision”.
-
Need
Plantix responds to the need of farmers with limited resources to have a reliable, fast, and cheap way to identify and treat crops diseases, pests, or nutrient deficiency. To make agriculture sustainable the grower must be able to make a profit.
-
Principle
The German company provides an App available on Google Store (iOS not available), that will help farmers in their fight against pests. When the farmer notice something wrong with one of his crops, he must take a close picture of it with his phone and send it to Plantix. He will almost immediately receive insights on the pest affecting his crop and how to treat it. The treatment can be pesticides or a local more environmentally friendly solution. The AI provides its results with <90% confidence every time. The model gets more accurate and learn to identify new diseases for every picture a user sends. In 2020, Plantix could recognize more than 500 different diseases and more than 10M photos were send to the servers. The App is free for farmers, but Plantix still make an income selling the data online with “Plantix Analytics”. A 99€ monthly subscription option is available on their website as well as a 15k€ per year package for big companies who wish to give the “Plantix Vision” tool to their employee. This tool comprises to regular Plantix App with added weather feature.
-
Main Technologies Involves
- User’s computer and smartphone
- Web coding languages: HTML/CSS/JS
- Electronic payment Solution
- Database
- Servers
- Machine Learning
- Android
-
Sources
- https://www.youtube.com/watch?v=7us0bTwNR18
- https://www.youtube.com/watch?v=0tQ__k3G17g
- https://medium.com/@savinavds/20-ai-and-tech-for-good-opportunities-in-food-agriculture-768f7eaaccc6
- https://plantix.net/en/
- https://plantix.net/en/analytics
- https://plantix.net/en/business
- https://www.eu-startups.com/2019/11/berlin-based-plantix-raises-e6-6-million-in-funding-to-further-improve-its-agricultural-app/
- https://plantix.net/fr/assets/custom/press-material/logos/plantix.png
-
B. Deepening
-
Plantix
-
Carriers and actors of the project
The main carrier of Plantix is the husband-wife duo Rob Strey and Simone Strey under their startup Progressive Environmental and Agricultural Technologies (PEAT). The company’s developers and data engineers are the main workforce of the project. The community that provides images to Plantix is also necessary to the project as it feeds the database and help the AI improve. Several German investors are supporting the project: RPT Capital, Piton Capital, Atlantic Lab, and Index Venture. Additionally, big companies buying Plantix other products (Plantix Analytics and Plantix Vision) make most of Plantix income. Without them Plantix would have next to no income and would not be sustainable.
-
Research question
How to help developing countries’ small farmers reduce crops loss and improve crops yields using AI?
-
The reason you selected this project
This project is a perfect example of how AI can help people in their daily life. The Plantix App is especially easy to use and only needs 3G to work in multiple local languages, the languages small farmers from developing are familiar with. There is even a WhatsApp chatbot in Indi whenever a user encounters an issue! Plantix App makes a real difference for people. It is a perfect fit for small farmers with low income who need fast and cheap answers to their problem to survive. It is for the concrete help that Plantix provides to small farmers that I chose this project.
-
-
User Scenario
-
Users
The main users are farmers with small parcels of land coming from developing countries and living in rural areas. Additionally, bigger companies use Plantix premium services, “Plantix Analytics” and “Plantix Vision”.
-
Persona
- Name: Shivay Kumar
- Age: 30yo
- Profession Wheat Farmer
- Place of residence India, Uttar Pradesh, countryside
- Family Status married with 3 children
- Income$4/day = INR300/day
- Hoobies writing, calligraphy, walking around his parcels and looking at the mountains
- Dreams and Desires working in illustration design, seeing his children take the farm when he is old
- Needs: earning more money to support his family
- Problems and Frustrations: not having enough time to practice calligraphy because his crops are too time consuming, spending money on useless pesticides, not being able to afford a good school in Mumbai to his youngest daughter (Aarna, 12 yo).
- Major Issue: his crops are dying from an unknown pest, but no one has any idea what it is and how to treat it. If he doesn’t do anything the harvest will be ruined.
-
Key Features
- Request a personalized diagnosis his crops
- Receive personalized treatment options for his crops
- Ask questions to the WhatsApp chat bot
- Post/read advice from the community section on the App
- Access subset of Plantix data
- Take a picture
- Send a picture to Plantix
-
UX Storyboard
-
-
Technical Analysis
-
General Principle
When a picture of a sick crop is sent to Plantix servers, it is processed by the AI which will analyze patterns to predict with over 90% confidence level the pest affecting the plant. The AI is trained on a dataset and is improving as users send new pictures
-
Technical Overview - AI version
The data is sent to the servers and analyzed by deep neural networks that will come up with a probability and a disease name. The AI is > 90% certain that this is the disease affecting the plant and gives the answer instantly. The AI uses classification to identify the patterns left by the pests or disease on the plant leaves. Colors, forms, shapes, regularity, … are analyzed.
Is it based on supervised or unsupervised learning?
Plantix is an image recognition tool, its algorithm is based on supervised learning. The labels are set by humans and the AI insert the imput under these labels.
What kind of data is manipulated?
Images with geolocation data and time data are manipulated in a very high volume.
What tasks does the algorithm have to performe?
The AI based on supervised learning uses classification mathods. To be exact, Plantix uses a multi-label, multi-class classification to recognize which disease affects the crop based on the value of the atributes.
Is it using datasets and how? What do these datasets contain, are data labelled, and if yes how? Where do the datasets come from?
The AI is using datasets created from the pictures the farmers send to Plantix to train the model. A received image creates a JSON response which will then feed the database. The data is labelled with crop name, size, … automatically when the picture is analyzed. This could mean that the dataset might contain a lot of noise as depending on the picture values in attributes can be misidentified, and the data can be mislabeled.
Is their some potential biases we have to be careful of?
There could be an eventual sampling bias, depending on the volume of pictures sent from people from specific regions. India provides the most pictures, and some regions in India are especially represented. Which means that pictures from other places are not equally likely to be selected in a sample. This could influence the algorithm. A way to counter this effect would be to give data from all regions an equal weight. However, some regions might not have enough data.
What are the types of algorithms used? In which language are they programmed, with which librairies?
The algorithms being used are property of PEAT Ghmb (Berlin). The programming languages used are shell, python, java and many more. PEAT uses Google’s TensorFlow software library for its image-recognition tool.
Sources:
https://papers.ssrn.com/sol3/Delivery.cfm/SSRN_ID3781281_code4478733.pdf?abstractid=3729753&mirid=1 https://docs.plantix.net/
https://www.forbes.com/sites/parmyolson/2018/10/15/this-startupbuilt-a-treasure-trove-of-crop-data-by-putting-ai-in-the-hands-of-indian-farmers/?sh=49df57931916 -
Added Value thanks to Artificial Intelligence
AI automates the process of identifying the diseases and pests. Without automation the process would require hundreds of experts looking at the pictures received from the farmers 24/7. This would not only cost an enormous amount of money but also would not be as accurate as the AI identification. Without the AI, the farmers would never have access to that kind of expertise.
-