NeuroSYS, Rybacka 7, 53-656, Wrocław, Poland.
Faculty of Fundamental Problems of Technology, Wroclaw University of Science and Technology, Wybrzeże S. Wyspiańskiego 27, 50-372, Wrocław, Poland.
Sci Rep. 2022 Mar 25;12(1):5212. doi: 10.1038/s41598-022-09264-z.
We introduce an effective strategy to generate an annotated synthetic dataset of microbiological images of Petri dishes that can be used to train deep learning models in a fully supervised fashion. The developed generator employs traditional computer vision algorithms together with a neural style transfer method for data augmentation. We show that the method is able to synthesize a dataset of realistic looking images that can be used to train a neural network model capable of localising, segmenting, and classifying five different microbial species. Our method requires significantly fewer resources to obtain a useful dataset than collecting and labeling a whole large set of real images with annotations. We show that starting with only 100 real images, we can generate data to train a detector that achieves comparable results (detection mAP [Formula: see text], and counting MAE [Formula: see text]) to the same detector but trained on a real, several dozen times bigger dataset (mAP [Formula: see text], MAE [Formula: see text]), containing over 7 k images. We prove the usefulness of the method in microbe detection and segmentation, but we expect that it is general and flexible and can also be applicable in other domains of science and industry to detect various objects.
我们提出了一种有效的策略，用于生成微生物培养皿的带注释的合成图像数据集，可用于以完全监督的方式训练深度学习模型。所开发的生成器采用传统的计算机视觉算法和神经风格迁移方法进行数据增强。我们表明，该方法能够合成看起来逼真的图像数据集，可用于训练能够定位、分割和分类五种不同微生物物种的神经网络模型。与收集和标记整个大型带注释的真实图像集相比，我们的方法需要更少的资源来获得有用的数据集。我们表明，仅从 100 张真实图像开始，我们就可以生成用于训练检测器的数据，该检测器的性能（检测 mAP [公式：见正文]和计数 MAE [公式：见正文]）与在更大的真实数据集（mAP [公式：见正文]，MAE [公式：见正文]）上训练的相同检测器相当，该数据集包含超过 7k 张图像。我们证明了该方法在微生物检测和分割中的有效性，但我们期望它具有通用性和灵活性，也可应用于科学和工业的其他领域，以检测各种物体。
