BeeGPT

Project description

The current bee identification process by taxonomists is time-consuming, not accurate enough, and with low usability. Based on this information, the main research question is "How can the bee identification process be improved by using AI solutions?". As each student specializes in a different field, the question is answered from multiple aspects.

Context

An estimated 1.6 trillion wild animals are accidentally killed by humans every year, a tragic consequence that underscores the importance of biodiversity conservation. One of the key tools in safeguarding biodiversity is taxonomy, the scientific discipline that classifies organisms into hierarchical categories based on shared characteristics. Accurate identification and classification of species are essential for implementing conservation strategies and mitigating the factors contributing to their decline.

Results

• Benchmark of five data-driven developed tools (LLMs, rule-based methods, RAG)
• Interactive maps following the trends and movement of bees throughout recorded history
• Designs, that improve the usability of the process, targeted at non-IT users
• Architecture layouts and diagrams that determine the best framework for dealing with sparse data.

About the project group

All students in the project have graduated a Bachelor in the field of IT. The duration of the project is 18 weeks and 40 hour working weeks have been followed throughout the course of the duration. Each student focuses on a different subtopic of the main goal - Improving the bee identification process of taxonomists by using AI solutions. By following a three week sprint schedule, students are able to deliver products in the fields of Organizational Processes, Software, Infrastructure, and UI/UX.