BEGIN:VCALENDAR
VERSION:2.0
PRODID:icalendar-ruby
CALSCALE:GREGORIAN
METHOD:PUBLISH
BEGIN:VEVENT
DTSTAMP:20260524T111821Z
UID:https://www.mps.mpg.de/events/45184/7840847
DTSTART:20260326T100000Z
DTEND:20260326T110000Z
CLASS:PUBLIC
CREATED:20260320T171959Z
DESCRIPTION: Recent advances in solar physics increasingly rely on automate
 d identification of coronal structures using machine learning. Yet most st
 udies emphasise scientific performance without evaluating feasibility for 
 onboard deployment to prioritise downlink observations. We investigate the
  automated identification of active regions and coronal holes by applying 
 segmentation and detection techniques to Solar Dynamics Observatory (SDO) 
 data. We compare three approaches: SCSS-Net\, a deep learning model for se
 mantic segmentation\; YOLOv8n\, a lightweight object detector\; and a trad
 itional pipeline based on basic computer vision operations (BCVO). Each me
 thod is assessed for its scientific accuracy and its suitability for deplo
 yment in future resource-limited missions. While no direct hardware benchm
 arking has been performed yet\, we assess the feasibility of an onboard im
 plementation based on the number of trainable parameters\, the architectur
 e\, and the associated hardware requirements. Training and evaluation are 
 first conducted on well-calibrated SDO images. We then extend the evaluati
 on to raw and uncalibrated SDO images affected by instrumental artefacts. 
 Performance is mainly measured using the Intersection over Union (IoU) and
  Dice score. Results show that while SCSS-Net achieves the highest segment
 ation quality\, YOLOv8n offers a strong balance between accuracy and effic
 iency. The BCVO pipeline remains viable under strict hardware limitations.
  Interestingly\, our models retain compatibility on Level-0 observations. 
 This is the first study comparing these widely used methods from the persp
 ective of onboard deployment. Our findings provide a foundation for design
 ing frameworks tailored to onboard hardware configurations.\nSpeaker: Pana
 giotis Gonidakis 
LAST-MODIFIED:20260320T172129Z
LOCATION:https://uio.zoom.us/j/63138938090
ORGANIZER;CN=Valeriia Liakh:mailto:
SUMMARY:ESPOS: ESPOS: Comparing Solar Structure Detection Methods in SDO/AI
 A Observations for Onboard Spacecraft Deployment (Panagiotis Gonidaki)
URL;VALUE=URI:https://www.mps.mpg.de/events/45184/7840847
END:VEVENT
END:VCALENDAR