BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//pretalx//talks.staging.osgeo.org//foss4g-2022//talk//DS9R78
BEGIN:VTIMEZONE
TZID:CET
BEGIN:STANDARD
DTSTART:20001029T040000
RRULE:FREQ=YEARLY;BYDAY=-1SU;BYMONTH=10
TZNAME:CET
TZOFFSETFROM:+0200
TZOFFSETTO:+0100
END:STANDARD
BEGIN:DAYLIGHT
DTSTART:20000326T030000
RRULE:FREQ=YEARLY;BYDAY=-1SU;BYMONTH=3
TZNAME:CEST
TZOFFSETFROM:+0100
TZOFFSETTO:+0200
END:DAYLIGHT
END:VTIMEZONE
BEGIN:VEVENT
UID:pretalx-foss4g-2022-DS9R78@talks.staging.osgeo.org
DTSTART;TZID=CET:20220824T144500
DTEND;TZID=CET:20220824T151500
DESCRIPTION:Object detection\, classification and semantic segmentation are
  ubiquitous and fundamental tasks in extracting\, interpreting and underst
 anding the information acquired by satellite imagery. Applications for loc
 ating and classifying man-made objects\, such as buildings\, roads\, aerop
 lanes\, and cars typically require Very High Resolution (VHR) imagery\, wi
 th spatial resolution ranging approximately from 0.3 m to 5 m. However\, s
 uch VHR imagery is generally proprietary and commercially available at a h
 igh cost. This prevents its uptake from the wider community\, in particula
 r when analysis at large scale is desired. HIECTOR (HIErarchical deteCTOR)
  tackles the problem of efficiently scaling object detection in satellite 
 imagery to large areas by leveraging the sparsity of such objects over the
  considered area-of-interest (AOI). This talk presents a hierarchical meth
 od for detection of man-made objects\, using multiple satellite image sour
 ces with different Ground Sample Distance (GSD). The detection is carried 
 out in a hierarchical fashion\, starting at the lowest resolution and proc
 eeding to the highest. Detections at each stage of the pyramid are used to
  request imagery and apply the detection at the next higher resolution\, t
 herefore reducing the amount of data required and processed. We evaluate H
 IECTOR for the task of building detection for a middle-eastern country\, e
 stimating oriented bounding boxes around each object of interest.\n\nFor t
 he detection of buildings\, HIECTOR is demonstrated using the following da
 ta sources: Sentinel-2 imagery with 10 m GSD\, Airbus SPOT imagery pan-sha
 rpened to 1.5 m pixel size and Airbus Pleiades imagery pan-sharpened to 0.
 5 m pixel size. Sentinel-2 imagery is openly available\, making their use 
 very cost efficient. The Single-Stage Rotation-Decoupled Detector (SSRDD) 
 algorithm is used. Given that single buildings are not discernible at 10 m
  GSD\, a bounding box does not describe a single building but rather a clu
 ster of buildings. The estimated bounding boxes at 10 m are joined and the
  resulting polygon area is used to further request SPOT imagery at the pan
 -sharpened pixels size of 1.5 m. In the case of SPOT imagery\, given the h
 igher spatial resolution\, one bounding box is estimated for each building
 . As a final step\, predictions are improved in areas with low confidence 
 by requesting Airbus Pleiades imagery at the pan-sharpened 0.5 m pixel siz
 e. Ablation studies show that HIECTOR achieves a mean Average Precision (m
 AP) score of 0.383 and 20-fold reduction in costs compared to using only V
 HR at the highest resolution\, which achieves a mAP of 0.452.\n\nCode will
  be released under MIT license. We will also release the trained models on
  Sentinel\, SPOT and Pleiades imagery. In addition\, manually labelled bui
 lding footprints over Dakar will be open-sourced to allow users evaluate t
 he generalisation of the models over different geographical areas. The Sen
 tinel Hub service is used by HIECTOR to request the commercial imagery sou
 rces on the specified polygons determined at each level of the pyramid\, a
 llowing to request\, access and process specific sub-parts of the AOI.
DTSTAMP:20260403T222549Z
LOCATION:Auditorium
SUMMARY:HIECTOR: Hierarchical object detector for cost-efficient detection 
 at scale - Devis Peressutti\, Nejc Vesel
URL:https://talks.staging.osgeo.org/foss4g-2022/talk/DS9R78/
END:VEVENT
END:VCALENDAR