How The Field Museum Unlocks New Research Possibilities with Vision AI

If you have millions of insect specimens, how do you digitize them and transform the imagery into useful research data?

In this interview with Elizabeth Postema, Postdoctoral Researcher at the Field Museum of Natural History, we explored how researchers from several institutions are using computer vision to digitize over 8 million insect specimens, turning them into valuable data for large-scale scientific research.

​Elizabeth shared insights into the development of DrawerDissect, a custom software solution that automates the transcription of specimen data and enables rapid analysis of insect features like color and size. She'll explain how they used Roboflow to build custom vision models and accelerated a process that would have been impossible with traditional methods.

Chapters
00:00 Introduction: Digitizing Insect Specimens with Vision AI
02:50 Why Museums Need to Digitize Collections
09:13 Challenges in Gathering Data to Study Morphology
13:28 Using Custom Vision AI to Speed Up Digitization
19:29 Measuring Insect Specimens with Segmentation
21:33 Text Detection and Transcription in Specimen Drawers
26:50 Using Data From Thousands of Specimens to Verify Bergmann's Rule
30:17 Imaging Specimens: 20 Years vs. A Few Weeks
35:22 Using Custom AI to Classify Species
38:04 Live Demo: Getting Started with DrawerDissect
44:09 Collecting Images and Calibrating Measurements
49:10 Journey from Researcher to AI Expert
54:12 Final Thoughts, Additional Audience Questions, and Future Direction

Join a future live session:
https://roboflow.com/webinar

Read the pre-print research paper here:
https://ecoevorxiv.org/repository/vie...