🤖 AI Summary
This study addresses the insufficient accuracy and robustness in adenovirus detection within transmission electron microscopy (TEM) images by systematically evaluating, for the first time, the performance of four advanced data augmentation strategies—NAS, GAS, GMAS, and DAS—across different YOLOv8 model scales. Under unified training conditions, the authors generated YOLO-compatible bounding boxes through refined re-annotation of adenovirus particle locations and established a standardized preprocessing pipeline. Experimental results demonstrate that specific augmentation techniques, particularly DAS, substantially improve detection accuracy. The study identifies the optimal combination of model architecture and augmentation strategy, offering an efficient and reliable technical pathway for automated virus particle detection in TEM imagery.
📝 Abstract
This study systematically benchmarks different data augmentation setups across YOLO26 model size variants to determine the most effective setup for adenovirus detection in TEM images. The benchmarked setups include NAS, GAS, GMAS and DAS, all evaluated under identical training conditions. The adenovirus dataset, selected from the published TEM virus dataset, was re-annotated by leveraging adenovirus particle positions to generate YOLO-compatible bounding box annotations. The experimental results demonstrated the impact of the benchmarked data augmentation setups on adenovirus detection with YOLO26 and indicated the most effective data augmentation setup.