Astronomy is a science the object of which is to study closely the origin of meteorids in the Universe, their entering the Earth’s atmosphere in the form of meteors and rocky objects on the Earth’s surface in the form of meteorites. The streaming of celestial objects in the Universe and their entering the earth’s atmosphere needs to be carefully and regularly observed for astronomic and security reasons. Although for the majority of people, meteors and other celestial objects entering the Earth’s atmosphere are an abstract notion, it is interesting to become familiar with their definitions, form, size and other important information available. Meteoric phenomena can be studied and researched by every individual, who is not closely linked to some astronomic organization. In contemporary astronomy, the importance of detecting meteors using the methods of artificial intelligence has been on constant increase. The reason being the fact that by way of computers, with the assistance of convolutional and artificial neural networks, the appearance of a meteor or another celestial object in the image can be detected in a high-quality manner. Computer detection method enables also a faster, higher quality and more precise form of detection than it was the case with detection methods used in the past, relieving the astronomers of extra work. The topic of this graduation thesis is the classification of images which contain meteor detection and those which do not, using convolutional neural networks. In order to show the application of artificial intelligence in the meteor detection process, Pyhton programming language was used in this graduation thesis, with a set of images inserted in the neural network. The results obtained for the set of 200, 400, 1000, 1400, 2000, 2600 images demonstrate how the operation and stability of the convolutional neural network are improved with the growth of FITS data which are given to the network for learning in the training process.