Wieland Ralf, Ukawa Chinatsu, Joschko Monika, Krolczyk Adrian, Fritsch Guido, Hildebrandt Thomas B, Schmidt Olaf, Filser Juliane, Jimenez Juan J
Leibniz Centre for Agricultural Landscape Research, Eberswalder Str. 84, 15374 Müncheberg, Germany.
Department of Food and Energy Systems Science, Graduate school of Bio-Applications and Systems Engineering, Tokyo University of Agriculture and Technology, Tokyo, Japan.
R Soc Open Sci. 2021 Mar 31;8(3):201275. doi: 10.1098/rsos.201275.
Soil samples from several European countries were scanned using medical computer tomography (CT) device and are now available as CT images. The analysis of these samples was carried out using deep learning methods. For this purpose, a VGG16 network was trained with the CT images (X). For the annotation (y) a new method for automated annotation, 'surrogate' learning, was introduced. The generated neural networks (NNs) were subjected to a detailed analysis. Among other things, transfer learning was used to check whether the NN can also be trained to other y-values. Visually, the NN was verified using a gradient-based class activation mapping (grad-CAM) algorithm. These analyses showed that the NN was able to generalize, i.e. to capture the spatial structure of the soil sample. Possible applications of the models are discussed.
来自几个欧洲国家的土壤样本使用医学计算机断层扫描（CT）设备进行了扫描，现在可以作为CT图像获取。这些样本的分析使用深度学习方法进行。为此，使用CT图像（X）训练了一个VGG16网络。对于注释（y），引入了一种新的自动注释方法，即“替代”学习。对生成的神经网络（NNs）进行了详细分析。其中，使用迁移学习来检查NN是否也可以针对其他y值进行训练。在视觉上，使用基于梯度的类激活映射（grad-CAM）算法对NN进行了验证。这些分析表明，NN能够进行泛化，即捕捉土壤样本的空间结构。讨论了模型的可能应用。
