BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//pretalx//talks.osgeo.org//foss4g-2026//speaker//NASYTF
BEGIN:VTIMEZONE
TZID:JST
BEGIN:STANDARD
DTSTART:20000101T000000
RRULE:FREQ=YEARLY;BYMONTH=1
TZNAME:JST
TZOFFSETFROM:+0900
TZOFFSETTO:+0900
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
UID:pretalx-foss4g-2026-FPNKWH@talks.osgeo.org
DTSTART;TZID=JST:20260901T153000
DTEND;TZID=JST:20260901T160000
DESCRIPTION:Abstract : \n\nBuilding footprint extraction from high-resoluti
 on satellite imagery requires accurate building boundary raster masks and 
 an effective shape reconstruction method to produce natural building footp
 rints. Unlike vertex-centric graph approaches or mask contour tracing\, we
  propose DINO-EdgeQuery\, an edge-first paradigm that addresses this need.
  A frozen DINOv3 backbone\, adapter neck\, and an instance decoder inspire
 d by Mask2Former provide building-aware boxes and masks\, while a region o
 f interest (ROI)-conditioned EdgeQuery decoder predicts edge activity\, ce
 nter\, direction\, length\, and successor relations. By capturing dominant
  wall structures as edge primitives and reconstructing vertices purely thr
 ough deterministic line intersections\, we deliberately separate neural ed
 ge prediction from verifiable geometric assembly. Active edges are ordered
  by successor scores\, and the P2 quality gate removes invalid or containe
 d duplicate polygons. This geometric post-processing encourages sharp\, li
 ne-intersection-based corners and enforces topology-safe polygons through 
 deterministic validity checks. To improve generalization\, we train segmen
 tation and EdgeQuery branches separately\, mask inactive losses to zero\, 
 and merge the trained weights for unified inference. Validation on the dat
 aset publicly released on the website by the Wuhan University Geospatial C
 omputer Vision Group demonstrates that our EdgeQuery framework\, including
  the P2 quality gate\, delivered high-quality\, geometrically superior pol
 ygon outputs with zero self-intersections and triangle collapses\, and sim
 ultaneously improved Intersection over Union (IoU) and suppressed duplicat
 e building footprints.
DTSTAMP:20260717T234857Z
LOCATION:Cosmos1
SUMMARY:DINO-EdgeQuery: Edge-First Polygon Decoding for Building Footprint 
 Extraction from Satellite Imagery - Yuji Kobayashi
URL:https://talks.osgeo.org/foss4g-2026/talk/FPNKWH/
END:VEVENT
BEGIN:VEVENT
UID:pretalx-foss4g-2026-3HSDCX@talks.osgeo.org
DTSTART;TZID=JST:20260903T130000
DTEND;TZID=JST:20260903T130500
DESCRIPTION:Orbitalnet discusses its journey toward realizing geospatial-In
 telligence and what lies ahead.
DTSTAMP:20260717T234857Z
LOCATION:Phoenix Hall
SUMMARY:Orbitalnet : Enabling Geospatial Intelligence Architecture - Yuji K
 obayashi
URL:https://talks.osgeo.org/foss4g-2026/talk/3HSDCX/
END:VEVENT
END:VCALENDAR