Neural Network Detection of Microparticles on Reconstructed Holographic Images of Volumetric Medium

Abstract—The article presents the results of using two neural networks for solving problems of detection and classification of microparticles in images obtained by reconstructing model and experimental holograms of aerosol media with a particle density of 180 000 cm${}^{-3}$. One neural network is trained on a database of manually labeled particle images and the second one is trained on a database created using algorithms for automatically classifying certain parts of images. The results of the work of two neural networks are compared and ways to improve them further are suggested/