Flower Power

Project description

Our project exists of 3 sub projects and thus has 3 different research questions.

Leslie has worked on researching and developing a digital image annotation.

Dimitar has worked on researching and designing an engaging mobile app that motivates citizen science volunteers to contribute to biodiversity conservation.

Sieuwe has worked on researching a new method of autonomous flight for drones using human flight data.

Context

Our project is developed as a collaboration to the ARISE-platform, an environment created and maintained by Naturalis for use in biodiversity monitoring.
Flower Power and our developed solutions will be integrated into the platform and add to the functionality which the platform can provide for it's users, which for now consists of data annotators, biologists, AI research and in the future citizen volunteers.

Results

As this project consists of multiple, seperate projects, we all had our own research questions and results.

Leslie's research regarding image annotation tools gives a better insight in User Interaction techniques and important features which help creating a new User Interface and improving the existing functionalities of Label Studio, to make it an easy to use annotation tool specialised for image annotation, which can be integrated into the ARISE platform.

Based on the research sieuwe has done into human flight data and drone autonomy, he concluded that human flight data could show improved generalization and reliability performance for obstacle avoidance in more complex environments. However training a CNN model to accurately represent the wide distribution of human flight behavior is challenging. Further research should be done into a loss function that can better handle wide distrubutions without losing accuracy on fine details.

After Dimitar conducted the research on how to motivate citizen science volunteers to contribute to biodiversity conservation, a mobile app was designed and developed. The app is designed to use the Flower Power AI model to analyze, identify and count wildflowers based on pictures uploaded by users. In addition to providing identification and counting results, the app also provides educational resources on wildflowers. The challenge was to design a user-friendly mobile app, with a simple and intuitive interface that allows users to easily upload pictures and receive results. The mobile app also includes features such as geolocation, user authentication, achievements and challenges, which enhance the user experience and promote engagement.

About the project group

All project group members are graduates of the HBO-ICT Bachelor course from Fontys.
Dimitar is interested in mobile app development, Leslie is interested in Gamedesign, software development and webdevelopment  and Sieuwe is interested in robotics.
We continued our studies with the Master of Applied IT and spend 40 hours a week working on the Flower Power project, divided over 5 days.

Video