iiinova — where your digital self keeps evolving

Reposted by Asger Svenning

Jamie Alison @jamiealison.bsky.social · 01/10/2026

Sensors and AI could transform how we track insect declines, but the road to robust “insect sensing systems” is riddled with pitfalls 🫗🐞🪰 @insectai.bsky.social & friends propose seven steps to dodge them, working responsibly with tech and AI to generate huge & standardized datasets Open access 1/

doi.org

Insect monitoring without pitfalls: Seven steps for robust insect sensing systems

Cameras, acoustics, radar and photonics allow us to observe insects at unprecedented spatial and temporal scales. Different sensors have unique strengths, while species ID is a critical bottleneck. ...

0127

Asger Svenning @asgersvenning.com · 21/10/2025

Oops, SVG illustrations are nice though! ;)

111

Asger Svenning @asgersvenning.com · 18/04/2025

Not entirely sure of the species, but definitely an umbellifer/Apiaceae of some kind. There are many great species for pollinators within it depending on where you live, such as wild carrot (D. carota) or Ammi majus (which the plant in the image may be) or many others.

010

Asger Svenning @asgersvenning.com · 14/04/2025

Dataset: doi.org/10.5281/zeno... Big thanks to my co-authors: Quentin Geissmann, Guillaume Mougeot, Toke T Høye, Juli Carrillo, Jamie Alison, Daphne Chevalier, Kim Bjerge, Nisa Molina, and Song-Quan Ong, as well as everyone else who contributed with data or annotations.

doi.org

flatbug-dataset a compilation of dataset of terrestrial arthropodes on various surfaces

This dataset is a compilation of 23 datasets all individually described in their respective DOI entry. The complete dataset was used to train "flatbug" a generalist instance segmentation algorithm tha...

120

Asger Svenning @asgersvenning.com · 14/04/2025

GitHub: github.com/darsa-group/... We also set a precedent for robust evaluation of machine learning methods for applications within not only biodiversity monitoring, but also ecology as a broader subject, pushing beyond simple averaged metrics and random mixing of training and testing data.

github.com

GitHub - darsa-group/flat-bug: Flatbug is a hyperinference and trained YOLOv8 model zoo, with a bespoke diverse dataset of the same name.

Flatbug is a hyperinference and trained YOLOv8 model zoo, with a bespoke diverse dataset of the same name. - GitHub - darsa-group/flat-bug: Flatbug is a hyperinference and trained YOLOv8 model zoo...

130

Asger Svenning @asgersvenning.com · 14/04/2025

Preprint: www.biorxiv.org/content/10.1... We introduce flatbug—an easy-to-use, versatile tool for detecting and segmenting insects and arthropods from many types of images. The tool aims to accelerate insect biodiversity monitoring, laboratory experiments and ecological studies.

biorxiv.org

A General Method for Detection and Segmentation of Terrestrial Arthropods in Images

To better understand the status and trends of insects and other arthropods, emerging technologies like image recognition are developing rapidly. This is creating a strong demand for efficient and accu...

110

Asger Svenning @asgersvenning.com · 14/04/2025

I am excited to share that my first first-author paper, “A General Method for Detection and Segmentation of Terrestrial Arthropods in Images,” is now available on preprint. If you are interested take a look at my blog (asgersvenning.com/flat-bug/) or consider reading the pre-print (linked below).

2359