Organised by:
in collaboration with:

secretariat:

 

social media:


 



Accesso Utenti

IAP - Institut d'Astrophysique de Paris
Address:
France
Miscellaneous Information:

Abstract Reference: 30889
Identifier: P1.22
Presentation: Poster presentation
Key Theme: 1 Reduction and Analysis Algorithms for Large Databases and Vice-versa

Deblending in crowded star fields using convolutional neural networks

Authors:
Paillassa Maxime, Bertin Emmanuel

Astronomical images with high stellar density pose a challenge to source extraction algorithms. We present a new star detection and deblending method based on a convolutional neural network. The obtained detector can deal with a wide variety of astronomical images, and shows notable gain in both completeness and reliability when compared to traditional single-pass algorithms. We discuss further prospects and improvements to the method.