The little things that run the world: What can computer vision learn from insect images?

Insects are everywhere around us, pollinating plants, recycling nutrients, controlling pests, and feeding other animals. Yet their small size and extraordinary diversity make insect communities difficult to monitor at large scales. Computer vision and deep learning offer new ways to identify and classify insects from images, raising a simple question: what more can we learn from an insect image?

During her PhD studies, Melika Baghooee investigated how insect images can be used both for species identification and to extract information beyond species identity. The thesis first examines whether body size information can improve image-based insect classification. It then introduces EntoScan, an open-source high-throughput method for imaging arthropod samples, and BEEomass, a deep learning model for estimating individual dry biomass of arthropods from images. Finally, these methods are combined and applied to field-collected arthropods from an agricultural experiment to examine how image-derived measures can support agroecological analyses.

The research contributes new tools and approaches for image-based arthropod studies, including size-aware classification, high-throughput imaging, individual biomass estimation, and the use of image-derived measures in an agroecological experiment.

The PhD study was completed at the Center for Quantitative Genetics and Genomics (QGG), Faculty of Technical Sciences, Aarhus University.

This summary was prepared by the PhD student.

Time: Friday, 25 September 2026 at 13:30.
Place: Building 1632, room 201, the AIAS Auditorium, Aarhus University, Høegh-Guldbergsgade 6B, 8000 Aarhus C
Title of PhD thesis: Insect Size Matters: Computer Vision for Insect Classification, Biomass Estimation, and Agroecological Research
Contact information: Melika Baghooee, e-mail: melika@qgg.au.dk, tel.: +45 9194 2904
Members of the assessment committee:
Professor Kim Steenstrup Pedersen, Department of Computer Science, University of Copenhagen, Denmark
Researcher Ignasi Bartomeus, Doñana Biological Station EBD-CSIC, Spain
Professor Doug Speed (chair), Center for Quantitative Genetics and Genomics (QGG), Aarhus University, Denmark
Main supervisor: Professor Luc Janss, Center for Quantitative Genetics and Genomics (QGG), Aarhus University, Denmark
Co-supervisor: Assistant Professor Quentin Geissmann, Center for Quantitative Genetics and Genomics (QGG), Aarhus University, Denmark
Language: The PhD dissertation will be defended in English

The defence is public.
The PhD thesis is available for reading at the Graduate School of Technical Sciences/GSTS, Ny Munkegade 120, building 1521, 8000 Aarhus C.