Neural Network Method for Removing the Effect of Atmospheric Noise in an Image

The purpose of this paper is to develop a neural network method for dealing with noise in the form of weather conditions. The existing methods for cleaning images from noise were analyzed. The architecture of neural network deraining method based on modernized ResNeXt is proposed. Based on the results of testing the combination of information from the detail layer and negative residual information, the training process is simplified by reducing the range of mapping. Experiments demonstrated successful transfer of the model to real-world images with a significant improvement in the prediction accuracy of the pre-trained model on photos from rainy conditions.