Every summer, LIST looks back at the success stories from the previous year. In the spotlight is Olivier Parisot who is exploring how Deep Learning approaches can help to produce clear astronomical images.
Today, Electronically Assisted Astronomy is widely used to observe deep-sky objects, such as nebulae or galaxies. By capturing images using a camera connected to an optical instrument, this approach aims to display enhanced views of the target objects in near real-time on a screen, by running a minimal image processing phase. Observing the night sky therefore becomes more accessible to the general public, and in particular to people who find it difficult to use a telescope directly, because of poor visual sharpness for example.
In this context and as part of the MILAN (MachIne Learning for AstroNomy) project, LIST and VAONIS, a French company specialising in the development and sale of a new generation of automatic and smart telescopes, are jointly exploring how recent Deep Learning approaches can help to produce clear and realistic images, even when observation conditions are not ideal. These innovative features should help the company to continue to raise its competitiveness and its unique positioning on the market.