InsectUp : my insect identifier app

On July 2017, I was leading a humanitarian mission in Cambodia. I used to work in rural areas, in the heart of nature. Between two shovels, I took photos of beautiful butterflies and dragonflies to make a pause. When I was a child, I knew all the species what we can find in France and I mapped them for the national database. In Cambodia, they were all unknown for me and it was a pleasure to identify them.

Photos I took with my mobile phone (Samsung S7)

But I wanted to report my observations and I discovered that there was no database in Cambodia. In addition to this, in Laos and Vietnam, the neighboring countries, twice as many different species as in Cambodia had been spotted. It was an anomaly since Cambodia presents the same type of environment. This gave me the idea of creating a world database of insects allowing everyone to contribute easily.

Voilà, my idea to create InsectUp was born. Let me introduce you to my product.

The insect population decline

First, let’s talk about an upcoming catastrophe : the decline of insect populations, particularly pollinating insects, is becoming a worldwide issue. Some studies report an 80% decrease in the number of insects seen in Europe in the last 30 years. Furthermore, researchers have not enough data to carefully study and monitor population decline because of difficulty to get reliable figures from small geographical areas. A first step to reverse the situation is to increase citizens awareness on that issue and mostly the young generations.

A first solution

We offer a mobile application and a web platform allowing automated recognition using artificial intelligence of photographed insects. The app highlights fun and educational approach to generate citizens awareness of mass disappearance of insects while collectively improving research database. To appeal users, an amazing insect picture hunting using the mobile phone camera has been developed to create an educative PokemonGo on actual species.

Our algorithm is now able to recognize 403 pollenating species in Eastern Europe with a success ratio of 87%, reaching a 95% for the most common ones such as bees or ladybugs.

Our algorithm

Based on a challenge hosted on the RampStudio platform, we were able to build a baseline for our classification algorithm. The SPIPOLL (Suivi Photographique des Insectes POLLinisateurs) proposes to quantitatively study pollinating insects in France. For this, they created a crowdsourcing platform where anyone can upload pictures of insects and identify their species through a series of questions. These data are then used by specialists for further analyses.

How it works. Credits : La Recherche, Septembre 2018

Our business model

We have three sources of incomes:
freemium and advertisements : we offer advanced features for experts and additional gamification features for fans ;
partnerships : with companies having now legal obligations to evaluate their environmental footprint, with local communities for which we are able to provide information on insect fauna, with local farmers to provide commercial visibility and potential marketing through our platform ;
sale of goodies : around the protection of insects — insect plants, insect houses, educational kits — and games for children accompanying the application and allowing their awakening.

The market

Public is young people for fun game, walker and trekker, gardeners, insect specialists who want to increase their. We also focus on educational markets (entomology, life sciences,…), zoos with butterfly aviaries, companies and farmers who have to carry out biodiversity monitoring.

An alpha version of the application was launched on playstore in April, 2018 exceeding 40K downloads, and now achieving 300 downloads per day and in more than 100 countries. This highlights the potential demand for our application. Source : Firebase

The team

We are a team of two Ecole Centrale Paris students in Computer Science Master degree who are passionate about insects, software development and artificial intelligence. We have the support of senior researchers from the Paris Saclay Center for Data science and the Natural History Museum of Paris who provide us databases, resources and advise to develop our algorithms.

Source: Deep Learning on Medium