Presentation

Invasive alien species (IAS) are plants and animals that are "introduced accidentally or deliberately into natural ecosystems where they are not normally found". They are characterized by being more resistant than the native species, they tolerate a wider range of environmental conditions, they have a short life cycle and have a prolific seed production and spread fast...etc. In other words, they exert a selection pressure on the latter and outcompete with them.
IAS have been identified as one of the major drivers of biodiversity loss. Currently, there are 66 IAS. But IAS surveillance, prevention and reduction are complex. First, IAS represent diverse species and ecosystems (30 are animals, 36 are plants, 1 is marine-species, 41 are primarily terrestrial species, 23 are primarily freshwater species..etc).2 Moreover, alien species risk assessment and classification into invasive species are time consuming and have a financial cost.
This is where Artifical Intelligence can enter into the equation and help achieve further this goal and reduce the IAS on land and water ecosystems.
In the first part of this case study, we will make an exploratory research, review and discuss four digital projects that solve the SDG issues that fall under our scope. We will detail the needs they answer to and the technologies employed to solve these issues.
In a second part of our analysis, we will go further visionning how the use of AI solves these problems and potentializes the solutions, the methodologies and tools used and if AI can provide with even further improvements.

Case Study : AI use & the Reduction of the impact of invasive alien species on land and water ecosystems.

    A. Exploratory research :

    I. Hawking Project:3,4


HAWKING project description. Source : https://business.esa.int/projects/hawking

  1. Project carriers :

    2. The project is carried by three entities. HummingBird technologies which is a world imagery analytics provider and who is the contractor and solution provider. There is the Soil and Agrifood Institute at Cranfield University who has expertise in soil management, soil conservation and ecosystem services. The least but not last important entity is Velcourt, an international farm management advisory group.
  2. Beneficiaries :

    3. The direct beneficiaries are any users in the agriculture supply value chain. But considering limited Earth resources, everyone in the world population will benefit from a more sustainable agriculture.
  3. Users :

    4. The users are any agriculture business owner or user, starting from farmers, to agronomists, land use regulators and environmental monitoring agencies, banks...etc
  4. Need :

    5. The project answers to the users'need to improve agriculture productivity, providing the users with crop growth models and accurate real-time information to manage and harvest their farmland sustainably.
  5. Principle :

    6. The Hawking project provides users with timely information about the weather, soil conditions, benchmarking performance of fields, crops health, rapid disease or weeds and pest damage identification...etc. The solution offered here is a real time accurate data to monitor the crops growth. It enables to monitor weather and soil conditions while identifying and treating any crop disease rapidly, mitigating agrochemical overuses and disease resistance build-up.
    The bottom line is that the users cut costs and improve their harvests while managing the land and water resources responsibly for a more sustainable agriculture.
  6. Main technologies involved :

    7. The system is a cloud-based web application using multiple technologies. We can list : the Hummingbird API, a thrid party API, drone imaging and satellite imaging, the Internet, servers, a deep learning cloud and satellite and UAV-mounted sensors (to monitor the atmospheric conditions).
  7. Sources :

    II. TSMF_10CM Project:5


  1. Project carriers :

    2. The TAMA group is the project contractor and is handling the forest real time imagery through automated image analysis.
  2. Beneficiaries :

    3. Beneficiaries of this project are forestry owners and operators who are now able to monitor stress to their forests.
  3. Users :

    4. The users are professional foresters managing about 500ha to 2,500ha of forests.
  4. Need :

    5. The project enables to detect early stress to forests due for instance to aggressive alien species proliferation and to take adapted management decisions to reduce damages.
  5. Principle :

    6. Stress detection is currently happening in a non-automatized monitoring, basically, a “walk and watch out”. Professional foresters need to tackle rapidly the stress issues and in a cost-effective way. The TSMF_10CM project brings them with an accurate timely stress map with the precise stress locations and degrees as well as with probable stress issues to be solved.
  6. Main technologies involved :

    7. The system is built on an online portal. It has two blocks, the first uses GNSS satellites which continuously transmit navigation signals to route the drones and geo-rectify the images. The second block is a monitoring module based on Earth Observation satellite (mainly Sentinel-2) and a ’Monitoring Pro’ module based on Unmanned Aerial Vehicles imagery with resolution of 10 cm.
  7. Source :

    III. VisioTerra Detection and Warning Service (VtDaws) Project:6,7


VtDaws Logo

  1. Project carriers :

    2. VtDaws is carried by VisioTerra, an independant French company specialized in Science Consulting for Earth Observation and by CBO consulting group, a strategy and funding for digital ventures.
  2. Beneficiaries :

    3. The beneficiaries are the end-users whatever the field considered, for example, in Agriculture, beneficiaries are the farmers, cooperatives, agrofood companies, in Water Management, it is water agencies and public bodies...etc
  3. Users :

    4. Target users are working in Agriculture (farmers, cooperatives, agrofood companies), Forestry (public and private), Oil&Gas (production companies, environmental NGOs), Water Management (water agencies, public bodies) and Insurance Sector (insurance companies, experts).
  4. Need :

    5. VtDaws system is implemented to detect changes or anomalies, and it sends warnings to the users.
  5. Principle :

    6. The applicability fields are numerous and multiple usages are possible. Regarding the SDGs in our scope, VtDaws can for example send pollutions or algae blooms warnings as soon as they are detected automatically in water bodies. For Agriculture, VtDaws enables the crop monitoring, for Forestry, VtDaws sends deforestation warnings, planting, pruning, cleaning, logging… and can also assess the impact of events such as fires, storms, hurricane or invasive species. VtDaws can also be used in the Water Management sector...etc
  6. Main technologies involved :

    7. The system uses an automated software system designed to provide clients with real-time monitoring system of their resources. It is built on an Earth Observation satellite and gathers other type of Data inlcuding meteorology, climate, Land Use/Land Cover, Digital Elevation Models and biogeophysical data. It is based on machine learning.
  7. Sources :

    IV. AGIIR :8


AGIIR contre les insectes ravageurs

  1. Project carriers :

    2. The solution is a public solution, provided by the INRAE, the "French National Research Institute for Agriculture, Food and the Environment". It is owned by the Republic of France.
  2. Beneficiaries :

    3. Beneficiaries belong to multiple species. First, there are the city inhabitants, the people living in the area and by extension the locality, the country (France) and the neighboring countries. Every human being is a beneficiary.
    Secondly, there are the plants, crops and any vegetable which is outcompeting with the IAS and for which its ecosystem is unbalanced.
    Final beneficiary is the animals.
  3. Users :

    4. The users are any of the french citizens or inhabitants. Any citizen can become a user and an actor.
  4. Need :

    5. The need is to do the right thing, to protect human beings, the Earth and other species and to preserve Earth for the future generations by becoming an actor in the collectivity against invasive alien species. It consists in identifying IAS, staying informed about them and supporting the authorities in controlling the dissemination of IAS that are harmful for public health, crops, plants or animals at a local and national french territory (France).
  5. Principle :

    6. The project is a bio-surveillance digital solution and decision making tool. It will enable mapping the IAS in France on a vountary basis through citizens.
    The solution is available in 4 languages (french, spanish, italian and english).
    AGIIR is available as a website or an application to download for free with a smartphone on Goople play or App Store. There is also a map* that should enable the mapping of the different IAS and enable the users to identify and search for IAS in their region.
  6. Main technologies involved :

    7. The project needs a smartphone to take a picture.
    Then the user needs the Internet, a website or an application where he will log in and upload the images he took. A server is needed to collect the data from the user. Then at the backoffice, the picture is validated (human intervention by the researchers) and the IAS confirmed (or not). The user is able to get the information available on the website about the IAS he identified.
  7. Sources :
  8. Other info and notes :

    9. The solution is available in two versions : website and downloadable application.
    * The map is not functionning properly on the website : the map reached the maximum size allowed for pictures. It is currently not available for the users.

    B. Deepening section :

    I. Selected project - AGIIR :8


AGIIR, project printscreen

  1. Carriers and actors :

    2. The project is carried by the INRAE, a public insitution with 18 research centers in France. The researchers working at the INRAE are the first actors. They are the ones validating the data collected through AGIIR and confirming the phenotype or status of the species identified by the citizens (users). They are the experts behind the scenes. The other main actors are the engaged citizens, France inhabitants, who are responsible actors of the civil society. They are doing the reporting through AGIIR at their individual scale.
  2. Research question :

    3. How can we motivate city residents to collect IAS data with their phones in order and contribute to public efforts made to map infested areas and enable an early detection of the newly introduced IAS ?
  3. The reason why I selected the project

    4. First of all, to achieve the SDG goals, I believe that we need to join forces and do it collectively, one step at a time, one person at a time.
    Within the SDGs, this project is close to me and impacts me. I come from Tunisia where palm trees are endangered. After going through AGIIR, I am able to understand the "WHY". I recognized two of the IAS described there. I would even like to get more information and see what I can do at an individual level to support the global efforts to reduce IAS and stop their expansion.
    As a citizen of the world I feel responsible. And I believe that this project serves the greater good and needs to be scaled up. AGIIR can reach its full potential first by designing a more user friendly website and application. It can also be powered by marchine learning enhancing both the user experience and increasing data collection.
    Currently, France is the only the scope. But I can see that people living in Belgium and Spain have gathered some data, which is now visible on the map. Nothing would prevent responsible citizens of the world to support data collection and stay informed about the IAS.
    I can also see another expansion of AGIIR that would benefit achieve reducing the IAS : there are 66 IAS in the UCIN IAS list but only 7 IAS are available and can be identifiable in AGIIR.
    To achieve the SDGs, all the citizens of the world need to get the rigth level of awareness and information so that to engage anyone in data collection for the public authorities and the research centers that can do something at their level to prevent and reduce IAS impact on the ecosystem (land and earth).

    II. User scenario :

  1. Users :

    2. AGIIR users are any France inhabitant equiped with a smartphone.
  2. Persona :

    3. Hi !
    My name is Régine,
    I am 65 years old, I was a school teacher. Now I am retired,
    I live in the “Côte d’Azur” with my husband.
    I enjoy taking my grand-kids to the parc. I love gardening.
    I eat ethical food when the vegetables don’t directly come from my garden !
    I believe in the French Republic and its values. I am a patriot.
    I also believe that we have a responsibility towards the future generations and that we should preserve our natural resources and the planet.
    I have a lot of free time, and I feel frustrated not to serve a purpose anymore.
    I am concerned about the future of the planet, about what we are leaving for the next generations.
    Currently, the major issue bothering me is :
    I found some insects in my garden, there are a few of them but they seem to be eating the wood ! And what if they also get inside the house ? I am concerned about this…

    Disclaimer : Régine is a persona, a fictional character presented for illustrative purpose.
  3. Key features :
    4. Take a picture
    Upload the image
    Identify the insect
    Get information about the IAS characteristics (at the user level)
    Share the IAS details information (with the researchers and the authorities)
    Get notified upon approval of the IAS data and a thank you message.
    Get the information and updates about what authorities are doing (not yet available but it could be a great feature)
  4. UX storyboard :
    5. UX Storyboard
  5. Technical Overview - AI version :

    6. In this section, we will discuss how artificial intelligence can be incorporated and its impacts on the user experience. With the 7 IAS species on their website, AGIIR has a dataset of images already classified. We have thus a good starting point to get an algorithm of image recognition. This model will be integrated to the website using the Javascript library ml5.js. This ml5.js library will use the model and a classification function to make predictions on the images that will be uploaded by the users.This is supervised learning.

    Using the library p5js which would also have been integrated to the website, the user will be able to drag and drop his pictures on AGIIR website. The process will consist on an html image id generation for every picture uploaded by the user. Each image will be classified in Javascript according to two variables, its ID (html image ID) and a second variable called a "classifier". The classifier will use the classification model (based on our existing dataset with pictures classified) and a function. This latter will generate predictions about the image to be classified.
    The probability of accuracy will also be stored in a new variable in Javascript. It is the level of confidence in the (IAS) classifaction prediction.
    As a consequence of the use of artifical intelligence on the website, the result (html) will be displayed immediately on the screen of the user in the form of a classification result and a level of confidence in the prediction.

    The user experience is much better than in the existin model of AGIIR.
    First, the user gets an immediate answer to his request (classification as an IAS or not). Second, he doesn't have to search by himself in the existing website or application documentation to get the information he needs. He gets the answer immediately after seeing the insect, he just needs to take his smartphone and a picture. That's it !

    We have also to consider the researchers aspect here. Artificial intelligence will also save time for the researchers who will not have to confirm the insect phenotype and its classification. The human scientific intervention would only happen in case of doubt.
    As a result of artifical intelligence all of the following will come instantaneously, in a fraction of seconds : data collection, IAS infestations mapping, identification of newly introduced IAS and users information.
    Isn't this beautiful ?

    C. Complementary sources :