Principal Investigators: Chris Biemann and Mariella Herberstein 

PhD student: Jingheng Pan

This project leverages cutting-edge AI-driven object recognition technologies to analyze morphological features in biological collections, focusing on insects and spiders. Advanced models like RAM (Recognize Anything Model) and SEEM (Segment Everything Everywhere Model) provide automatic, high-precision identification of traits such as size, color, hairiness, and more, without predefined categories. The goal is to apply these tools to museum and collection data to track morphological changes over time and across environmental conditions, especially relating to climate change.

Employing a multi-modal approach, the project combines images, metadata, and environmental data to better understand how climate-related factors influence phenotypic traits. By comparing historical and current collections, the project aims to develop a systematic understanding of morphological evolution driven by environmental shifts. The resulting analysis platform will be highly generalizable, applicable across various taxa and collections.

The research work is divided into three main phases. First, it involves meticulous data collection and annotation from scientific collections, systematically gathering and structuring both old and new data. Next, the collected data will be analyzed using specialized object recognition models, which will be fine-tuned and further developed for optimal performance. In the final phase, a systematic multi-modal analysis of morphological changes over time and between populations will be performed, including the development of automated procedures to detect extreme variations and patterns.

This innovative platform will significantly support other projects focused on trait analysis and museomics by providing a versatile toolbox for comparative morphological studies. Ultimately, it aims to generate new insights into how climate change impacts animal morphology and to further develop the methods for broader applications in digital morphometrics and phenotypic research.

Foto: Jingheng Pan (AI (Gemini) generated)