Tails: Chasing Comets with the Zwicky Transient Facility and Deep Learning

May 2021 • 2021AJ....161..218D

Authors • Duev, Dmitry A. • Bolin, Bryce T. • Graham, Matthew J. • Kelley, Michael S. P. • Mahabal, Ashish • Bellm, Eric C. • Coughlin, Michael W. • Dekany, Richard • Helou, George • Kulkarni, Shrinivas R. • Masci, Frank J. • Prince, Thomas A. • Riddle, Reed • Soumagnac, Maayane T. • van der Walt, Stéfan J.

Abstract • We present Tails, an open-source deep-learning framework for the identification and localization of comets in the image data of the Zwicky Transient Facility (ZTF), a robotic optical time-domain survey currently in operation at the Palomar Observatory in California, USA. Tails employs a custom EfficientDet-based architecture and is capable of finding comets in single images in near real time, rather than requiring multiple epochs as with traditional methods. The system achieves state-of-the-art performance with 99% recall, a 0.01% false-positive rate, and a 1-2 pixel rms error in the predicted position. We report the initial results of the Tails efficiency evaluation in a production setting on the data of the ZTF Twilight survey, including the first AI-assisted discovery of a comet (C/2020 T2) and the recovery of a comet (P/2016 J3 = P/2021 A3).

Links

• PDF https://iopscience.iop.org/article/10.3847/1538-3881/abea7b/pdf
• PDF https://stacks.iop.org/1538-3881/161/218/pdf
• PREPRINT http://arxiv.org/abs/2102.13352
• ELECTR https://doi.org/10.3847%2F1538-3881%2Fabea7b