Master Project, 2022
▪ MSE Project, ~ 360h
▪ Based on the Project Mitwelten
▪ Advisor: Thomas Amberg
Human existence depends on biodiversity. Pollinators play a special role because plant reproduction depends on them. Due to human actions and environmental influences, the population of pollinators varies greatly around the world. Since they play a key role in biodiversity, monitoring pollinators is important to detect changes early on. Within the scope of this project, an automated evaluation was developed for systematically collected image recordings. The evaluation pipeline consists of two steps, where flowers are detected in the first step and pollinators are detected on the flowers in the second step. Five classes of pollinators can be detected. The objects are reliably recognized, making scientifically relevant statements about biodiversity feasible. Four concepts for automated evaluation were developed and tested. Three of them are carried out on edge devices, one in the backend. The smallest system is based on a single board computer and can be operated offline for long periods of time. The most efficient version is the evaluation in the backend.
Open documentation ↗