Anomaly Detection in Astronomical Objects of Galaxies Using Deep Learning

Space exploration has been an area of study that has been probed since a very long time, and with the improvements in technology over time, the techniques and approaches have evolved and so have the amount and the nature of data received. The astronomical data we receive these days has increased to the extent that it is physically impossible to rely only on human effort to analyze all that data, as there is always new data being generated every day in large amounts. Aided with machine learning techniques, in our project, we explore different challenges faced by astronomers dealing with this huge data and look for relevant information according to each objective. Firstly, we propose to use Inception v3 for classification and Fast RCNN for image segmentation to detect and classify galaxies. These approaches contribute a great deal in certain areas of research that have massive amounts of redundant data to be discarded according to the need of the study with the help of good performance of Python in the domain that studies processing of images and Computer vision. Using this we can have a better look at relevant observations and can reach suitable conclusions accordingly using computational power, which reduces the need of researchers to go through each and every bit of data obtained from space telescopes, sky surveys, etc.