Deep learning for space exploration


With the recent advancement in the image acquisition technology, several lunar and other planetary explorations have provided a massive amount of planetary image data to the research community. However, it is very difficult for humans to analyze such huge amount of image data manually for further scrutiny. Therefore, in this research we employ deep learning techniques to analyze planetary images in a lot faster and automated way. We propose deep learning algorithms to classify areas of scientific-interest such as central-peak craters on the lunar surface. We also solve the image restoration task on Lunar/Mars surface image by intelligently predicting the surface image in a context-aware fashion to fill in the missing pixels so as to obtain the entire surface image.