Speaker
Description
Two parallel trends have spurred the development of automated insect monitoring technology: increasing awareness of insect population declines and growing capabilities of artificial intelligence. LEPMON is a collaborative project of five German institutes aiming to design and test an AI-driven system for species-level monitoring of moths and other nocturnal insects. The project encompasses hardware development, data management, AI identification, citizen science for trap operation and image annotation, and ecological field testing. LEPMON started operation in 2025, and here I will present its first results. We have deployed our Automatic Recorders of Nocturnal Insects at 47 locations, along 5 urbanization gradients and recorded well over 1 million images, corresponding to 1.1 million bounding boxes. We are currently in the process of annotating the images to train the AI.
Effective monitoring requires recording and communicating reliable numbers of individuals per species. To investigate if automatic recorders derive such reliable numbers, we tested how often individual moths are recorded on the screen, and thus how high the chances of double counts are. We did this by a mark-recapture study of 31 individual moths in 2025, and found that almost all moths visited the screen multiple times per night, and that a non-negligible number of moths (5 out of 31) were seen on more than one night. When counting the number of individuals of each species, the likelihood of double counts is thus very high, and we are currently developing statistical procedures to correct these multiple counts post-hoc.
| Status Group | Senior Scientist |
|---|---|
| FOR TALKS: Poster Presentation Option | Undecided/No preference |