BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//CERN//INDICO//EN
BEGIN:VEVENT
SUMMARY:Developing a Recurrent U-Net for Faint Trans-Neptunian Object Dete
 ction Using Synthetic Injections
DTSTART;VALUE=DATE-TIME:20261028T175500Z
DTEND;VALUE=DATE-TIME:20261028T180500Z
DTSTAMP;VALUE=DATE-TIME:20261028T201734Z
UID:indico-contribution-3642@nyx.physics.mcgill.ca
DESCRIPTION:Speakers: Malavika Nair (Western University)\nTrans-Neptunian 
 Objects (TNOs) are small\, distant bodies beyond Neptune that provide impo
 rtant context on the formation and evolution of the outer Solar System. Ho
 wever\, the faintest TNOs are difficult to detect because their weak signa
 ls can be easily confused with noise and subtraction artifacts. Manual ide
 ntification of genuine TNOs is therefore a time-consuming process that can
  require hours of careful inspection. To automate the process\, this proje
 ct develops a recurrent U-Net framework for identifying faint moving sourc
 es\, with particular emphasis on generating realistic and well-characteriz
 ed training data. A synthetic injection pipeline\, maketensor\, was develo
 ped to implant artificial TNOs directly into real difference-image observa
 tions. Each source is propagated across multiple exposures with physically
  motivated apparent motion and injected using the point-spread function co
 rresponding to each observation. The underlying noise\, stellar residuals\
 , detector features\, and other observational artifacts are preserved to k
 eep the training data authentic. The pipeline also reproduces characterist
 ic subtraction effects and negative wells\, allowing the synthetic objects
  to more closely resemble real TNO signatures. The resulting multi-frame t
 ensors contain real observational backgrounds with exact ground-truth loca
 tions and trajectories for every injected source. These sequences are used
  to train a recurrent U-Net to produce pixel-wise probability maps that id
 entify likely TNO locations across a sequence of exposures. By learning th
 e consistent motion of a TNO across successive frames\, the network aims t
 o distinguish genuine moving sources from noise and image-subtraction resi
 duals\, enabling faster and more reliable detection of faint TNOs in large
  astronomical datasets.\n\nhttps://nyx.physics.mcgill.ca/event/587/contrib
 utions/3642/
LOCATION:Online 
URL:https://nyx.physics.mcgill.ca/event/587/contributions/3642/
END:VEVENT
END:VCALENDAR