EVALUATION OF THE VIDEOS WITH ARTIFICIAL INTELLIGENCE
Currently, several teams from ETH and the University of Zurich are working on optimizing the algorithm.
Using Deep Learning for Non-Invasive Pollen Detection on Honey Bees
by Nicholas Dykeman Data Science Lab ETH Zurich
Dominique Heyn Data Science Lab ETH Zurich
In this work, we present a completely non-invasive system for counting bee and pollen traffic in a beehive. This system is designed to be generalizable to various beehives, rather than being restricted to certain backgrounds and entrance types. In addition, we present a prototype cloud-based system that allows beekeepers to upload video footage, automatically processes this, and returns gathered statistics to the beekeepers. We examine the performance of our models and compare it to human performance, which we in many cases match or even outperform.