BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//pretalx//talks.staging.osgeo.org//foss4g-2022//speaker//3SSH9P
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-2022-8NGRBD@talks.staging.osgeo.org
DTSTART;TZID=CET:20220824T141500
DTEND;TZID=CET:20220824T144500
DESCRIPTION:A few years after the democratization of semantic segmentation 
 advanced techniques in the 2D\ncontext (imagery)\, there are more and more
  initiatives for exploiting such algorithms with 3D\ndatasets. The context
  appears favorable: public and private initiatives are arising in terms of
 \nmassive 3D dataset collection\, hence a huge amount of 3D point cloud da
 ta will become available in\na near future. As an example\, the french car
 tography institute (IGN) is currently targetting a\nfull-coverage of the c
 ountry with LIDAR data in the five next years.\n\nConsidering 3D point clo
 uds is a really challenging task regarding semantic segmentation. Whilst\n
 this data format allows to represent a scene with a high level of details\
 , the unordered and\nunstructured nature of the data makes the standard co
 nvolution neural network approach\nineffective. However other deep learnin
 g algorithms exist to cope with these\ncharacteristics. Depending on the d
 esired accuracy and the labelled data availability\, some\n"softer" machin
 e learning approaches may also complete the toolbox.\n\nLeveraging georefe
 renced data in such a context may be an interesting avenue in order to imp
 rove\nthe algorithm performances. In any case\, these fairly innovative so
 lutions can be applied in some\ngeographical use cases\, e.g. cartography 
 with street-views\, Building Information Modelling (BIM)\,\n...\n\nThis pr
 esentation will provide some insights on these 3D semantic segmentation re
 lated topics:\n\n* the 3D semantic segmentation state-of-the-art will be f
 lied over\;\n\n* the BIM use case will be detailed through the presentatio
 n of an ongoing R&D project carried out\nby Bimadata.io\, Oslandia and the
  LIRIS lab (CNRS)\;\n\n* the geo3dfeatures project (https://gitlab.com/Osl
 andia/geo3dfeatures) will be showcased in\norder to illustrate what the se
 minal component of a 3D point cloud segmentation software could\nbe.
DTSTAMP:20260403T204306Z
LOCATION:General online
SUMMARY:On the road to 3D semantic segmentation - Raphaël Delhome
URL:https://talks.staging.osgeo.org/foss4g-2022/talk/8NGRBD/
END:VEVENT
END:VCALENDAR