Connect a million nature observers to the species that surround them

The iNaturalist species and data sharing platform has reached a milestone earlier this month with its one millionth observer. The platform consists of a mobile application and a matching website that helps users identify the plants and animals around them. The tools allow users to upload photos and information about these sightings and to get help identifying them, within this nature-oriented community.

The platform aims to connect people to the nature around them, to encourage the sharing and evaluation of data and to provide scientists with data from the crowd that would otherwise be difficult to generate. Several technologies in smartphones, including camera, clock, calendar and GPS, help to tag each photo temporally and geographically, which allows iNaturalist data to contribute to information on the range of species .

Launched 10 years ago, the platform has received and processed nearly 14 million sightings from more than 180,000 species. He encourages observers to explore the outdoors by helping them identify what they see around them regularly, as well as at events such as the City Nature Challenge held earlier this year. .

The platform has since added an artificial intelligence to its toolbox, which has helped identify more common species. It relies on a large database of photos provided by the community. The numerous images, with the confirmed identifications it contains for certain species, have enabled the platform's collaborators to form computer vision algorithms to identify the species from the images, which allows the application to propose species names to observers in later uploaded photos.

The platform is also expanding to what it calls the iNaturalist network of institutions and partner sites in different countries. These sites, known as gateways, work together and share infrastructure, while personalizing the naturalistic community of each country to make local nature more accessible and to facilitate the availability of biodiversity data for regional science and conservation. The network currently includes bridges in the United States (iNaturalist.org), Mexico (NaturaLista), New Zealand (iNaturalist NZ – Mātaki Taiao), Canada (iNaturalist Canada), Colombia (Naturalista) and Portugal ( Biodiversity4All).

Mongabay interviewed Scott Loarie, co-founder and director of iNaturalist, on some of the technological changes observed in the development of the platform.

MB: What changes in technology or user base have surprised you the most?

Loarie: Advances in computer vision technology have surprised me the most. A few years ago, the idea of ​​forming a model to immediately identify a photograph with one of the tens of thousands of species seemed very exaggerated. But it's also counterintuitive to see how these models work. Some people think they are replacing naturalists, but it's really much more than a synthetic expertise within the iNaturalist community. The model only works as well as it has been formed.

MB: What additions do iNaturalist or its observers want to see in the near future?

Loarie: We are very interested in improving the quality of the computer vision model. This will require not only technological advances, but also an extension of the scope and expertise of the naturalistic community, as well as new ways to mobilize the community to collect the data needed for the community. improvement of the model.

MB: How the technologies behind the site, such as artificial intelligence and communications, has evolved since the creation of the site?

LoarieINaturalist has always wanted to know if we could create a community of people capable of transforming photos of living beings into high quality data for science and if we could evolve this "scientific instrument".

Technology has certainly helped to solve some problems. For example, previously, this machine was fully outsourced, which meant that users had to wait from a few hours to a few weeks to get an ID on their photos. Computer vision technology has streamlined this process for a large number of photos, which helps the system to evolve, but it has also introduced other concerns such as how to maintain high data quality.

MB: How have technological advances increased the risk for data quality?

Loarie: In the past, if someone typed identifying the Latin name on a photo, it was likely that they knew what they were doing, so the quality on that scale was high. More and more, people are adding more and more identifications suggested by the computer vision algorithm, so that even an amateur can produce a scientific name.

We are monitoring the ecosystem to see how these changes affect the quality of the data generated by the system while doing a lot of research and development on more rigorous methods of data quality.

MB: So it seems that the improvements brought by the new technologies have not necessarily been a linear process.

Loarie: New technologies have allowed us to meet the challenges and expand, but each step forward introduces new challenges. It's a bit like evolution where we hope we'll have a better "eye" over time, but it needs to work at every stage of evolution. So we have to make sure that our changes are not so dramatic that they break the system. .

FEEDBACK: Use this form to send a message to the publisher of this post. If you want to post a public comment, you can do it at the bottom of the page.