Efficient Brood Cell Detection in Layer Trap Nests for Bees and Wasps: Balancing Labeling Effort and Species Coverage
Researchers have developed a deep learning approach to automate the detection and classification of brood cells in layer trap nests used for studying wild bees and wasps. This method aims to reduce the significant manual labor involved in analyzing these nests, which often contain densely packed cells and exhibit class imbalance between common and rare species. A novel Constrained False Positive Loss (CFPL) strategy was introduced to mitigate the impact of unlabeled data and improve model performance while balancing accuracy and labeling effort. AI
IMPACT Automates a tedious manual process in ecological research, potentially accelerating biodiversity studies.