| Abstract [eng] |
The aim of this work is to construct and apply a convolutional neural network to classify multi-band images of the M51 galaxy, based on whether the frame contains a stellar cluster and found cluster’s parameters. 20000 artificial images are generated for training and testing the neural network. The network is trained successfully and manages to reach 73% accuracy in classifying artificial data. 85% of M51 clusters from a catalogue by Hwang and Lee (2008) are classified as clusters. Clusters display a correlation between their catalogue-based FWHM and neural network-determined σ. |
