BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//pretalx//talks.staging.osgeo.org//foss4g-europe-2025//speaker//9S
 7SHC
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-europe-2025-YG7B8H@talks.staging.osgeo.org
DTSTART;TZID=CET:20250717T151500
DTEND;TZID=CET:20250717T152000
DESCRIPTION:Automatic and reliable 3D point cloud classification is a cruci
 al yet challenging task with applications across various domains\, includi
 ng urban planning\, 3D modeling\, and the development of smart cities. Air
 borne LiDAR (Light Detection and Ranging) has emerged as an efficient and 
 effective tool for conducting large-scale 3D surveys of urban areas\, offe
 ring high spatial resolution and accurate data collection. Over the years\
 , numerous algorithms and methodologies have been proposed for point cloud
  classification. Despite advancements in machine learning and deep learnin
 g\, this task remains a significant challenge in the geospatial community.
  \nOne of the primary challenges lies in the availability of sufficient la
 belled data for training classification algorithms. The creation of public
 ly accessible\, large-scale datasets is essential for developing and bench
 marking new methods. While several databases have been introduced\, such a
 s ISPRS Vaihingen (Niemeyer\, et. al.\, 2014)\, LASDU (Ye\, et. al.\, 2020
 ) or AHN3 (AHN\, 2024)\, they have some limitations. For instance\, while 
 LASDU and AHN3 datasets are valuable resources for point cloud classificat
 ion\, they lack the comprehensive diversity of urban-specific classes\, li
 miting their utility in capturing the complexity of dense urban environmen
 ts. \nThe ISPRS benchmark dataset is most commonly used in resources in th
 is field. It provides a point cloud classified into nine classes\, along w
 ith features such as x\, y\, z\, intensity\, return number\, and the numbe
 r of returns. Additionally\, the synchronized orthophoto of the same area 
 is available\, providing valuable context for classification tasks. Howeve
 r\, this dataset also presents some challenges\, including its highly unba
 lanced class distribution and the relatively small number of points availa
 ble\, particularly for training deep learning methods. \n\nIn this paper\,
  we introduced an aerial LiDAR point cloud dataset\, UNS Novi Sad\, design
 ed specifically for the classification of complex urban environments. The 
 dataset comprises over five million points\, classified into seven distinc
 t classes\, and is focused on the City of Novi Sad\, which is known for it
 s unique urban morphology. The city’s layout reflects the architectural 
 and planning styles typical of Southeastern Europe in the post-World War I
 I era\, featuring a mix of high-density residential blocks\, green spaces\
 , wide boulevards\, narrow streets\, and diverse building types. These cha
 racteristics ensure that the dataset captures a wide range of structural a
 nd spatial variations.  In addition to providing new data\, we evaluate th
 e PointNet and PointNet ++ algorithm for classification of the proposed da
 taset.\n\nThe UNS Geo dataset comprises over five million points\, classif
 ied into 8 distinct classes\, and is focused on the City of Novi Sad\, whi
 ch is known for its unique urban morphology. The city’s layout reflects 
 the architectural and planning styles typical of Southeastern Europe in th
 e post-World War II era\, featuring a mix of high-density residential bloc
 ks\, green spaces\, wide boulevards\, narrow streets\, and diverse buildin
 g types. These characteristics ensure that the dataset captures a wide ran
 ge of structural and spatial variations. \nThe study area is in the urban 
 area of Novi Sad\, consisting of Liman\, located in the southeast part of 
 the city\, and the left Danube bank with high residential blocks\, spaciou
 s green areas\, and boulevards. The topography of the study area is flat\,
  with an average elevation of 77 m. The ALS point cloud data were collecte
 d using a Riegl LMS-Q680i laser scanner and a digital camera DigiCam H39 o
 nboard a helicopter. \nThe total number of annotated points is 5.4 million
  of points. The dataset is divided into two .las files: for training and f
 or testing. \nIn the .las file\, each point was assigned the following att
 ributes: Position: X\, Y\, Z coordinates of each point in UTM 34N (EPSG:32
 634) projection\, Intensity\, Return number\, Number of returns\, Classifi
 cation\, Scan Angle Rank\, Time\, RGB: Each \nRegarding the labeling\, the
  automatic\, semi-automatic\, and manual classification was used. We selec
 ted classes with a focus on different applications such as mapping\, urban
  planning\, and forestry monitoring. The points are classified into 8 diff
 erent classes: ground\, roads\, parking\, pedestrian lens\, roof\, walls\,
  high vegetation\, and cars. The training and testing datasets have a simi
 lar distribution\, except for pedestrian lenses and cars. The high vegetat
 ion class points contain the largest number of points. This is expected si
 nce the multiple returns are characteristic of this class and it is also a
  commonly occurring class in this type of city. The car class only reaches
   2.64 % of all labeled points\, making them one of the most challenging c
 lasses to detect. The imbalance of classes should be considered during the
  training or testing phase.\n\nTo provide a brief evaluation of the propos
 ed dataset\, the supervised classification to label points was performed. 
 The PointNet architecture is a neural network that directly classifies raw
  point cloud. PointNet++ applies PointNet to local neighbourhoods to captu
 re local features. The evaluation metrics include the recall\, precision\,
  and F1 score.
DTSTAMP:20260527T015040Z
LOCATION:PA01 (Quarticle)
SUMMARY:UNS Geo: LiDAR Dataset for point cloud classification in urban area
 s - Gordana Jakovljevic
URL:https://talks.staging.osgeo.org/foss4g-europe-2025/talk/YG7B8H/
END:VEVENT
END:VCALENDAR