BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//pretalx//talks.staging.osgeo.org//foss4g-2022//talk//PJDH8K
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-PJDH8K@talks.staging.osgeo.org
DTSTART;TZID=CET:20220826T141500
DTEND;TZID=CET:20220826T144500
DESCRIPTION:Building footprint extraction is a popular and booming research
  field. Annually\, several research papers are published showing deep lear
 ning semantic segmentation-based methods to perform this kind of automated
  feature extraction. Unfortunately\, many of those papers do not have open
 -source implementations for public usage\, making it difficult for other r
 esearchers to access those implementations.\n\nHaving that in mind\, we pr
 esent DeepLearningTools and pytorch_segmentation_models_trainer. Both are 
 openly available implementations of deep learning-based semantic segmentat
 ion. This way\, we seek to strengthen the scientific community sharing our
  implementations.\n\nDeepLearningTools is a QGIS plugin that enables build
 ing and visualizing masks from vector data. Moreover\, it allows the usage
  of inference web services published by pytorch_segmentation_models_traine
 r\, creating a more feasible way for QGIS users to train Deep Learning Mod
 els.\n\npytorch_segmentation_models_trainer (pytorch-smt) is a Python fram
 ework built with PyTorch\, PyTorch-Lightning\, Hydra\, segmentation_models
 .pytorch\, rasterio\, and shapely. This implementation enables using YAML 
 files to perform segmentation mask building\, model training\, and inferen
 ce. In particular\, it ships pre-trained models for building footprint ext
 raction and post-processing implementations to obtain clean geometries. In
  addition\, one can deploy an inference service built using FastAPI and us
 e it in either web-based applications or a QGIS plugin like DeepLearningTo
 ols.\n\nResNet-101 U-Net Frame Field\, ResNet-101 DeepLabV3+ Frame Field\,
  HRNet W48 OCR Frame Field\, Modified PolyMapper (ModPolyMapper)\, and Pol
 ygonRNN are some of the models available in pytorch-smt. These models were
  trained using the Brazilian Army Geographic Service Building Dataset (BAG
 S Dataset)\, a newly available dataset built using aerial imagery from the
  Brazilian States of Rio Grande do Sul and Santa Catarina. Pytorch-smt als
 o enables training object detection and instance segmentation tasks using 
 concise training configuration.\n\nThis way\, considering the aforemention
 ed\, this talk presents the usage overview of both technologies and some d
 emonstrations. Using metrics like precision\, recall\, and F1\, we assess 
 the results achieved by the implementations developed as a product of our 
 research\, showing that they have the potential to produce vector data mor
 e efficiently than manual acquisition methods.\n\nDeepLearningTools is ava
 ilable at the QGIS plugin repository\, while pytorch_segmentation_models_t
 rainer is available at Python Package Manager (pip). The Brazilian Army Ge
 ographic Service develops both solutions\, making their codes available at
  https://github.com/phborba/DeepLearningTools and https://github.com/phbor
 ba/pytorch_segmentation_models_trainer.
DTSTAMP:20260403T193014Z
LOCATION:Room 4
SUMMARY:Building Footprint Extraction in Vector Format Using pytorch_segmen
 tation_models_trainer\, QGIS Plugin DeepLearningTools and The Brazilian Ar
 my Geographic Service Building Dataset - Philipe Borba
URL:https://talks.staging.osgeo.org/foss4g-2022/talk/PJDH8K/
END:VEVENT
END:VCALENDAR