Counting Bees with Deep Learning
18. March 2021.
Based on the challenge "EnBeeMo", students of HM’s Department of Computer Science and Mathematics were given the task to improve the method to count bees developed in the previous year. For this, they used the method of Deep Learning. After a theoretical introduction to the topic "Deep Learning" as well as some exercises, the project and the problem were introduced by Prof. David Spieler as an application scenario. The aim of the challenge was also to investigate the influence of hyperparameters on the recognition performance in this scenario in more detail. The proposed solutions were presented in the form of Python notebooks and one paper per team. In the future, these could be used to improve the optimization process and to develop a recognition model for bees as performant and as real-time capable as possible. This would support the hardware developed by EnBeeMo (picture). During the challenge, the students received support from AWS and the DTLab.