Аннотация:This study introduces a method for the quantitative analysis of cross-section scanning electron micrographs of graphene oxide membranes. This method involves the segmentation of micrographs using a neural network built on the UNet architecture that was trained on synthetic data. Subsequently, the OrientLayer algorithm is applied to calculate the orientational order parameters of lamellae forming a membrane. The lamellae ordering was compared with the ordering of the oxidized graphene planes determined by using the spin probe technique. High prospects for the joint application of scanning electron microscopy and electron paramagnetic resonance techniques to study the internal structure of graphene oxide membranes were discussed.