Butterflies are sensitive bioindicators of environmental change. That means their numbers and diversity tell us a lot about the health of our environment. When butterfly populations fall, it often signals bigger problems like habitat loss, land-use changes, or climate impacts.

In Germany, large-scale butterfly research has been pioneered by the Helmholtz Centre for Environmental Research (UFZ) within the Helmholtz Association. Since 2005, UFZ and its partners have coordinated the Tagfalter-Monitoring Deutschland (Butterfly Monitoring Germany, TMD) project.

In this project, citizen scientists walk fixed routes and record butterfly species. Over more than twenty years, they’ve collected millions of observations. The results are clear, while some generalist species remain stable, many specialist species are in decline. This shows the urgent need for continuous biodiversity monitoring.

The Problems

Butterfly monitoring is valuable but often limited to trained volunteers. Existing apps can be clunky, hard to use, or miss the point. Instead of making it easy to take a photo, they overwhelm users with complicated steps. As a result, many people who might contribute simply don’t.

From our persona research, we see that most people still focus on their families and daily routines. Some enjoy nature, but it’s not their main priority. If an app feels difficult or demands too much effort, they won’t keep using it. What they need is a tool that matches their habits: simple, direct, and rewarding. Without this, valuable observations are lost, and biodiversity monitoring suffers.

Existing apps for butterfly monitoring already exist, but many are difficult to use, packed with confusing features, or not very engaging. This turns away the very people who could contribute the most.

That’s why we’re developing a new tool called ButterflyWatch, designed to be simple, fun, and accessible for everyone.

Target Group

Everyone can use ButterflyWatch, but we also focus on specific groups:

• Retired People → spend time outdoors, stay active, and contribute to science in a simple, rewarding way.

• Students/Teachers → learn science by doing, not just reading, and also can be used as interactive lesson.

• Nature lovers → Simply enjoys and loves nature.

• Researchers and conservation groups → receive more reliable, large-scale data to track trends.

The broader the group, the stronger the data and the bigger the impact.

The Solutions

ButterflyWatch is an app that makes butterfly spotting as simple as taking a photo

Tap the big camera button. Point it at a butterfly. The app’s AI checks the image, suggests the species, and records the location and time automatically. You earn points for each valid sighting, complete quests, and redeem coupons from local partners.

Inside the app, you’ll find your own collection, like Pokédex from Pokémon game, showing all the species, you’ve spotted. At the same time, the data you provide contributes to biodiversity monitoring that scientists can use.

 

The Prototype

We built a first prototype to show how ButterflyWatch will work in practice. The design is simple and familiar, so anyone can use it without a tutorial.

• The camera button is large and centered at the bottom. It’s the main action.
• Camera recognition uses AI to detect butterflies and confirm the sighting before awarding points.
• My Collection displays spotted species in a grid view, with locked silhouettes for species you haven’t found yet.
• My Path lets you record walks and see which butterflies you found along your route.
• Quests and Rewards motivate you to keep going. Complete challenges, gain points, and redeem coupons.

The AI plays a central role in making the prototype work. It supports species identification by detecting and classifying butterflies from photos or live video, helps with gamification by assigning points automatically, ensures reliability through spam and fraud detection, and improves data quality by filtering out blurry or duplicate images.

We also thought about how AI is used responsibly. To protect your data, images are processed locally and only metadata such as species, location, and timestamp are saved. Faces or unrelated objects in the background are blurred automatically before anything is uploaded. Every user has an equal chance, since the AI only checks for butterfly presence and liveness, not the quality of the camera. To stay transparent, the app displays confidence scores, for example: “80% sure: Peacock Butterfly.” While the model may perform better on common species than rare ones at first, we plan to retrain it continuously with diverse datasets, including community contributions, to reduce bias.

Technically, the prototype relies on a model called YOLOv8, which runs directly on the device. The app captures multiple pictures, rates their quality, and suggests the best one for submission. For those who want more control, a manual mode allows you to reevaluate and choose the final image.

This prototype isn’t the final version, but it demonstrates how ButterflyWatch will look and feel. The focus is on clarity, big buttons, and one action per screen so children, adults, and even older people can join without frustration.

 

Our Team

Team 7

• Anton
• Erik
• Joy
• Manon
• Pascal
• Penny
• Teddy

 

Appendix: