GeoML challenge

The GeoML Challenge is a web application developed by the data team at ERNI Switzerland. It allows users to playfully test what they know about road traffic against an artificial intelligence.

Developing an interactive challenge with OpenData

The ERNI Switzerland data team developed the GeoML Challenge in a very short time. This interactive challenge tests the user’s ability to assess traffic. For this purpose, the web application shows them aerial images of road traffic, and they then assess the hazard potential of the situation. The user’s own categorization of the hazard is then compared to an assessment by an artificial intelligence.

"The GeoML Challenge showed us that something great can happen when different ERNI practice areas work together on a project."

To develop the challenge, publicly available data from multiple sources was used to train a neural network. By cleverly combining what are essentially simple data sets, such as satellite images of a road, it was possible to draw conclusions about the potential for accidents in a particular traffic situation.

The GeoML team

All members of the data team were excited about the idea of implementing such a challenge and making it available to the general public. After an extensive brainstorming phase, the first prototypes were built, and the team then worked with the UX, marketing, and software engineering teams at ERNI Switzerland to develop the implementation strategy. Through an extensive testing period, the ever-growing team optimized the interactive challenge. The GeoML Challenge has been launched and is being used by many interested participants to test their own assessment skills. 

"ERNI has supported our project from the very beginning. Whether it was our direct supervisors, people from other practice areas, or teams from other countries, everyone was there when we needed help,"