Deep Neural Networks in Medical Microbiology for Bacterial Colonies Classification: Curated Dataset (COCO and YOLO)

This deposit contains a frozen snapshot of the “Micro Colonies” curated dataset exported from Roboflow in both COCO and YOLOv8 formats. The dataset includes 165 agar-plate images with 1,801 annotated bacterial colonies across three bacterial species, organized into predefined train, validation, and test splits. COCO annotations are provided as JSON files located at "train/_annotations.coco.json", "valid/_annotations.coco.json", "test/_annotations.coco.json". YOLOv8 format is provided with images under "{train,valid,test}/images/" and label files under "{train,valid,test}/labels/", with dataset configuration in data.yaml This snapshot is intended to support reproducibility of the experiments reported in the associated manuscript on deep learning based bacterial colony detection and classification.本存档包含从Roboflow导出的、采用COCO（Common Objects in Context）与YOLOv8两种格式的“Micro Colonies”精选数据集的冻结快照。该数据集包含165张琼脂平板图像，共标注了3种细菌的1801个细菌菌落，并按照预设划分为训练集、验证集与测试集。 COCO格式标注以JSON文件形式提供，存放路径分别为"train/_annotations.coco.json"、"valid/_annotations.coco.json"、"test/_annotations.coco.json"。 YOLOv8格式数据集的图像文件存放于"{train,valid,test}/images/"路径下，标签文件存放于"{train,valid,test}/labels/"路径下，数据集配置文件为data.yaml。 本快照旨在支撑关联论文中所报道的基于深度学习的细菌菌落检测与分类实验的可复现性。