The detection of medical equipment in healthcare facilities presents significant operational challenges, particularly when assets are partially obscured, impacting inventory tracking and resource management efficiency. Traditional detection systems demonstrate performance limitations when encountering occluded medical assets, which requires improved detection methodologies. This investigation examines the implementation of YOLOv8, an advanced object detection model, for identifying medical equipment under various occlusion conditions in hospital environments. The study utilized a systematic evaluation approach across multiple medical equipment categories, incorporating controlled occlusion patterns ranging from 30% to 50% of visible equipment features. Experimental analysis revealed that the YOLOv8 model achieved detection metrics of 0.938 (F1-score and accuracy) for unobstructed equipment, while maintaining performance levels of 0.889 (F1-score) and 0.867 (accuracy) under occlusion conditions. The model's confidence metrics demonstrated systematic adaptation to visibility constraints, with values decreasing from 0.896 to 0.760 in occluded scenarios. These quantitative results indicate that YOLOv8 effectively maintains detection capabilities under partial occlusion conditions while providing appropriate uncertainty quantification. The findings contribute to the advancement of automated medical asset detection systems, though further validation across diverse clinical environments is warranted to establish comprehensive operational reliability.
