SlideShare una empresa de Scribd logo
1 de 54
Descargar para leer sin conexión
FHNW COLLOQUIUM
March 16, 2021
SWISS TERRITORIAL DATA LAB
APPLIED DATA SCIENCE
Raphael Rollier - Nils Hamel – Huriel Reichel
Adrian Meyer - Christian Dettwiler
 Introduction about the Swiss Territorial Data Lab
 4D Platform
 Register of Buildings and Dwellings
 Objects Detection
 A Cantonal perspective
 Q&A
Program
A co-creation
project
A way to share
and replicate
A space for
experimentation
Swiss Territorial Data Lab (STDL) is...
If you want to go fast, go alone. If you want
to go far, go together
STDL partners are…
Solving concrete problems in public
administrations with Geo Data Science
STDL mission is…
What is the most effective way to find out the construction
date of buildings to complete the Building Register ?
How can I detect automatically changes in the field
in order to update the Land Register more rapidly ?
How can I improve the monitoring of solar energy
usage by detecting automatically panel installations ?
How can I monitor more effectively the
development of mining ?
The type of challenges we are exploring…
FHNW COLLOQUIUM
March 16, 2021
THE TIME DIMENSION
RBD COMPLETION RESEARCH PROJECT
Nils Hamel – Huriel Reichel
STDL 4D PLATFORM
SRTM EXAMPLE – 470 GB (ASC)
2009-10 2013-04 2017-04
THE TIME DIMENSION
EXAMPLE OF GENEVA LIDAR – 250 GB (LAS)
THURGAU – 2020-10-17
INTERLIS – ITF
THURGAU – 2020-10-13
INTERLIS – ITF
TEMPORAL DERIVATIVE
2020-10-13 & 2020-10-17
REGISTER OF BUILDINGS & DWELLINGS
Federal Statistical Office (OFS/BFS) & STDL
●
Federal Register
●
Missing buildings construction years
●
Automating construction years gathering
Two complementary research approaches
PROJECT
NATIONAL MAPS
Using swiss 1:25’000 national maps
●
Tracking the buildings on the maps
●
Detection of their appearence
●
Covering 2020 to 1950
APPROACH
SWISS NATIONAL MAPS
HOMOGENEOUS AND STABLE METHODOLOGY
Emergence of the notion of 3D raster
2010 2004 1998 1993
DEDUCTION PROCESS
DETECTIONS & MORPHOLOGICAL CRITERION
Simple case
Complex case
VALIDATION METRIC
RESULTS ASSESSMENT
Manually gathered sets of synchronous buildings
Register : 1962 – National Maps : 1960-1964
RESULTS ASSESSMENT
NATIONAL MAPS APPROACH
With an average temporal resolution of 5.8 years : 84.7%
STATISTICAL
Using statistical urban model
●
Workaround the lack of maps
●
Improve construction years approximation
●
Covering years before 1950
APPROACH
THEORETICAL BACKGROUND
BURGESS URBAN MODEL
Pattern in urban growth & spatial dependence
METHODOLOGY
FILLING THE GAPS IN THE DATABASE
Temporal variance to compute spatial radii
METHODOLOGY
SPATIO-TEMPORAL CLUSTERING
Approximation of urban model : segment in ranges
1953
1955 1959
RESULTS ASSESSMENT
STATISTICAL APPROACH
95% of building correctly placed within a 31 years interval
●
National Maps Approach
84.7% within ±5.8 years
●
Statistical Approach
95 % within ±31 years
●
The importance of Time Dimension
Provides relevant information
CONCLUSION
FHNW COLLOQUIUM
March 16, 2021
OBJECT DETECTION FRAMEWORK
Adrian Meyer
STDL
OBJECT
DETECTION
FRAMEWORK
Generating a Model from
Cadastral Vectors and
Aerial Images
Predicting Objects in the
Same or a New Area of Interest
TRAINING
with Known Objects
INFERENCE
for Unknown Objects
BASIC IDEA
WORKFLOW
TRANSFER
LEARNING
Massive
Training
Dataset A
Acquired
Knowledge
Deep Learning
System
B(A)
Small Specific
Dataset B
Exported from
Cadastre
Deep Learning
System
A
New Predictions
Source: https://cdn-media-1.freecodecamp.org/images/1*lMEd6AcDmpH0mDzBHyiERw.png
Source: https://medium.com/swlh/object-detection-and-instance-segmentation-a-detailed-overview-94ca109274f2
Area of Interest AoI
Training AoI with Pool Labels
Prediction AoI
ZONING PLAN
Canton Thurgau
LAYER
WATER BASINS
Cadastre Export
Exclusion of Industrial Areas
AoI Tiling
Legend
Cantonal Boundary
AoI Boundary
Labels to be checked
Outlier Labels (discarded)
TRAINING
Ground Truth Generation:
Dataset Evaluation Split
80% Training
Used to Train Model Weights
10% Validation
Tuning Model Parameters
10% Test
Unbiased Assessment
Ground Truth Labels
7
DEEP LEARNING
RESULTS
How Good Did We Do?
# True Positives
# False Negatives
# False Positives
Registered &
Undetected (FN)
Detected, but
not registered (FP)
Registered &
Detected (TP)
Wikimedia Commons, 2021
F1 Score
The F1 Score is the Harmonic Mean of Precision and Recall.
Wikimedia Commons (2021)
P. Mirla (2018), via github.io
Metrics
Test Dataset at Zoom Level 19
Trained on SWISSIMAGE from Geneva and Neuchatel
Precision and Recall are a Function of the Confidence
84%
Thurgau Predictions:
Cadastre Updates
True Positives
• Threshold THR ≥ 5%
• 2‘227 of 2‘959 Pools detected
• 75% Detection Success
 25% (732) «wrongly» in
Cadastre ?
Thurgau Predictions:
Cadastre Updates
False Negatives
• 732 of 2‘959 Segments
>5m² missed
 25% «wrongly» in cadastre?
Thurgau Predictions:
Cadastre Updates
False Positives
• Threshold THR ≥ 97%
• 271 non-listed Pools
 9% missing in Cadastre?
• Threshold THR ≥ 94%
• 672 non-listed Pools
 23% missing in Cadastre?
Manual
Evaluation
Frauenfeld
Visible Pools
Dataset Metrics
True Positive: 81
False Positive: 9
False Negative: 18
F1-Score: 85.7%
Precision: 90.0%
Recall: 81.8%
- - - - - - - - - - - -
Detector Metrics
True Positive: 94
False Positive: 16
False Negative: 5
F1-Score: 90.0%
Precision: 85.5%
Recall: 94.9%
Swimming Pools: Zoom Level Results
Zoom Level 15
≈ 480 cm/px GSD
16
≈ 240 cm/px GSD
17
≈ 120 cm/px GSD
18
≈ 60 cm/px GSD
19*
≈ 30 cm/px GSD
File System Load
1’949 Tiles
= 0.4 GB
4’216 Tiles
= 1.3 GB
12’817 Tiles
= 4.0 GB
42’990 Tiles
= 11 GB
154’861 Tiles
= 40 GB
Processing Duration
(Prep. / DL+Pred. / Postp.)
± 25 min ± 40 min ± 120 min = 2 h ± 240 min = 4 h ± 1’200 min = 20 h
Max. F1 Score on
TST Dataset
55.5 % 75.3 % 82.5 % 83.4 % 84.1 %
* Zoom Level 20 contains 583’014 Tiles = 179 GB and was expected to take ~120h, but aborted prematurely due to bandwidth memory errors
What’s Next for the
End Users?
Update Cadastre
Send Letters to Swimming Pool Owners
Think about Detectable Objects and
Contact Us!
Silage Bale Detector
Labeling Strategy
- Manually Digitizing 200 Stacks of
Silage Bales in QGIS
- Training a Preliminary Detector
- Use 300 Highest-Confidence
Predictions as New Labels
- Manual Correction and Filling In of
Complete Training Tiles
Result
700 Labels in 1.5 Days
Wikimedia Commons (2021)
New Elements: Silage Bales
Zoom Level 16
≈ 240 cm/px GSD
17
≈ 120 cm/px GSD
18
≈ 60 cm/px GSD
19
≈ 30 cm/px GSD
20
≈ 15 cm/px GSD
File System Load
8’000 Tiles
= 1.3 GB
25’000 Tiles
= 8.0 GB
84’000 Tiles
= 26 GB
310’000 Tiles
= 80 GB
1’310’000 Tiles
= 320 GB
Processing Duration
(Prep. / DL+Pred. / Postp.)
± 40 min ± 120 min = 2 h ± 240 min = 4 h ± 900 min = 15 h ± 6’000 min = 100 h
Max. F1 Score on
TST Dataset
52.5 % 74.7 % 87.2 % 92.3 % 90.9 %
Silage Bale Detector
Silage Bale Detector
SWISS TERRITORIAL DATA LAB
APPLIED DATA SCIENCE
Sicht eines Kantons: Thurgau
(Einbezogen in die Projekte 4D, Schwimmbäder, Ko-
Produktion Hoheitsgrenzen, Siloballen)
A Cantonal perspective
Potenziale aus Sicht eines Kantons: 4D-Plattform
 Die amtliche Vermessung ist Referenzdatensatz!
Änderungen wirken sich auf viele «aufbauende
Geodaten» aus. => Information wertvoll
 Datenmigration auf neue IT, neues Datenmodell:
Bleiben die Daten korrekt und vollständig?
 Bauverwaltungen: Entsprechen die Änderungen
den neu bewilligten Bauten? Gibt es nicht bewilligte
Bauten?
SECTION
A Cantonal perspective
Potenziale aus Sicht eines Kantons: Objektdetektion
 Schwimmbäder: Im TG eher sekundär, als Test der
KI-Tools wertvoll
 Es gibt viele andere Potenziale: Bsp: Siloballen,
Echo des Landwirtschaftsamtes: Begeisterung
Die Objektdetektion sollte nicht auf das Thema
amtliche Vermessung eingeschränkt werden. Die
Kunden haben vielfältige Interessen, diese sollten
wir «abholen». Das stärkt unsere Position.
SECTION
A Cantonal perspective
Potenziale aus Sicht eines Kantons: STDL = Toolbox
 Die STDL-Tools können mit etwas Phantasie und
vor allem mit Kenntnis der Kundenbedürfnisse breit
eingesetzt werden.
Leitsatz Amt für Geoinformation Thurgau:
Wir schaffen mit Geoinformation volkswirtschaft-
lichen Nutzen.
Die STDL-Toolbox hilft uns dabei.
SECTION
FHNW COLLOQUIUM
March 16, 2021
SWISS TERRITORIAL DATA LAB
APPLIED DATA SCIENCE
Nils Hamel – Huriel Reichel – Adrian Meyer
Raphael Rollier – Christian Dettwiler
info@stdl.ch
stdl.ch

Más contenido relacionado

La actualidad más candente

Earth Observation and applications on environmental studies
Earth Observation and applications on environmental studiesEarth Observation and applications on environmental studies
Earth Observation and applications on environmental studiesICGCat
 
GIS WORKSHOP 18.11.2015
GIS WORKSHOP 18.11.2015GIS WORKSHOP 18.11.2015
GIS WORKSHOP 18.11.2015yllferizi
 
MIS 08 Geographical Information System
MIS 08  Geographical Information SystemMIS 08  Geographical Information System
MIS 08 Geographical Information SystemTushar B Kute
 
FME and the BGS in 2016/2017
FME and the BGS in 2016/2017FME and the BGS in 2016/2017
FME and the BGS in 2016/2017Sterling Geo
 
2nd e-ROSA Stakeholder Workshop: By, EO Based Global Public goods
2nd e-ROSA Stakeholder Workshop: By, EO Based Global Public goods2nd e-ROSA Stakeholder Workshop: By, EO Based Global Public goods
2nd e-ROSA Stakeholder Workshop: By, EO Based Global Public goodse-ROSA
 
Using satellite imagery to track economic change
Using satellite imagery to track economic changeUsing satellite imagery to track economic change
Using satellite imagery to track economic changeRishabh Srivastava
 
Geographic Information Systems (October – 2017) [Question Paper | CBSGS: 75:2...
Geographic Information Systems (October – 2017) [Question Paper | CBSGS: 75:2...Geographic Information Systems (October – 2017) [Question Paper | CBSGS: 75:2...
Geographic Information Systems (October – 2017) [Question Paper | CBSGS: 75:2...Mumbai B.Sc.IT Study
 
Amin tayyebi: Big Data and Land Use Change Science
Amin tayyebi: Big Data and Land Use Change ScienceAmin tayyebi: Big Data and Land Use Change Science
Amin tayyebi: Big Data and Land Use Change Scienceknowdiff
 
The Application of GIS in Urban Planning
The Application of GIS in Urban PlanningThe Application of GIS in Urban Planning
The Application of GIS in Urban Planningagungwah
 
nCOVID-19 pivot-and-fan map
nCOVID-19 pivot-and-fan mapnCOVID-19 pivot-and-fan map
nCOVID-19 pivot-and-fan mapAndrew Zolnai
 
Redesign of map.geo.admin.ch - 2013
Redesign of map.geo.admin.ch - 2013Redesign of map.geo.admin.ch - 2013
Redesign of map.geo.admin.ch - 2013Moullet
 
East Anglia Fenlands
East Anglia FenlandsEast Anglia Fenlands
East Anglia FenlandsAndrew Zolnai
 
The benefits of forecasting icing on wind energy produc- tion Øyvind Byrkjeda...
The benefits of forecasting icing on wind energy produc- tion Øyvind Byrkjeda...The benefits of forecasting icing on wind energy produc- tion Øyvind Byrkjeda...
The benefits of forecasting icing on wind energy produc- tion Øyvind Byrkjeda...Winterwind
 
Winning Ways for your Visualization Plays by Mark Grundland
Winning Ways for your Visualization Plays by Mark GrundlandWinning Ways for your Visualization Plays by Mark Grundland
Winning Ways for your Visualization Plays by Mark GrundlandPyData
 

La actualidad más candente (19)

Earth Observation and applications on environmental studies
Earth Observation and applications on environmental studiesEarth Observation and applications on environmental studies
Earth Observation and applications on environmental studies
 
Basics to gis concepts unit i
Basics to gis concepts unit iBasics to gis concepts unit i
Basics to gis concepts unit i
 
GIS WORKSHOP 18.11.2015
GIS WORKSHOP 18.11.2015GIS WORKSHOP 18.11.2015
GIS WORKSHOP 18.11.2015
 
MIS 08 Geographical Information System
MIS 08  Geographical Information SystemMIS 08  Geographical Information System
MIS 08 Geographical Information System
 
CCDUG - Dr. Peter Haas
CCDUG - Dr. Peter HaasCCDUG - Dr. Peter Haas
CCDUG - Dr. Peter Haas
 
FME and the BGS in 2016/2017
FME and the BGS in 2016/2017FME and the BGS in 2016/2017
FME and the BGS in 2016/2017
 
2nd e-ROSA Stakeholder Workshop: By, EO Based Global Public goods
2nd e-ROSA Stakeholder Workshop: By, EO Based Global Public goods2nd e-ROSA Stakeholder Workshop: By, EO Based Global Public goods
2nd e-ROSA Stakeholder Workshop: By, EO Based Global Public goods
 
Using satellite imagery to track economic change
Using satellite imagery to track economic changeUsing satellite imagery to track economic change
Using satellite imagery to track economic change
 
GIS
GISGIS
GIS
 
Geographic Information Systems (October – 2017) [Question Paper | CBSGS: 75:2...
Geographic Information Systems (October – 2017) [Question Paper | CBSGS: 75:2...Geographic Information Systems (October – 2017) [Question Paper | CBSGS: 75:2...
Geographic Information Systems (October – 2017) [Question Paper | CBSGS: 75:2...
 
Amin tayyebi: Big Data and Land Use Change Science
Amin tayyebi: Big Data and Land Use Change ScienceAmin tayyebi: Big Data and Land Use Change Science
Amin tayyebi: Big Data and Land Use Change Science
 
The Application of GIS in Urban Planning
The Application of GIS in Urban PlanningThe Application of GIS in Urban Planning
The Application of GIS in Urban Planning
 
nCOVID-19 pivot-and-fan map
nCOVID-19 pivot-and-fan mapnCOVID-19 pivot-and-fan map
nCOVID-19 pivot-and-fan map
 
Redesign of map.geo.admin.ch - 2013
Redesign of map.geo.admin.ch - 2013Redesign of map.geo.admin.ch - 2013
Redesign of map.geo.admin.ch - 2013
 
East Anglia Fenlands
East Anglia FenlandsEast Anglia Fenlands
East Anglia Fenlands
 
Fundamentals of gis
Fundamentals of gisFundamentals of gis
Fundamentals of gis
 
The benefits of forecasting icing on wind energy produc- tion Øyvind Byrkjeda...
The benefits of forecasting icing on wind energy produc- tion Øyvind Byrkjeda...The benefits of forecasting icing on wind energy produc- tion Øyvind Byrkjeda...
The benefits of forecasting icing on wind energy produc- tion Øyvind Byrkjeda...
 
application of gis
application of gisapplication of gis
application of gis
 
Winning Ways for your Visualization Plays by Mark Grundland
Winning Ways for your Visualization Plays by Mark GrundlandWinning Ways for your Visualization Plays by Mark Grundland
Winning Ways for your Visualization Plays by Mark Grundland
 

Similar a Swiss Territorial Data Lab - geo Data Science - colloque FHNW

LINUX Tag 2008: 4D Data Visualisation and Quality Control
LINUX Tag 2008: 4D Data Visualisation and Quality ControlLINUX Tag 2008: 4D Data Visualisation and Quality Control
LINUX Tag 2008: 4D Data Visualisation and Quality ControlPeter Löwe
 
New information sources for rain fields
New information sources for rain fieldsNew information sources for rain fields
New information sources for rain fieldsAndreas Scheidegger
 
BA Summit 2014 Predictive maintenance: Met big data het lek dichten
BA Summit 2014  Predictive maintenance: Met big data het lek dichtenBA Summit 2014  Predictive maintenance: Met big data het lek dichten
BA Summit 2014 Predictive maintenance: Met big data het lek dichtenDaniel Westzaan
 
Development of a soil carbon map for the United Republic of Tanzania
Development of a soil carbon map for the United Republic of TanzaniaDevelopment of a soil carbon map for the United Republic of Tanzania
Development of a soil carbon map for the United Republic of TanzaniaExternalEvents
 
NDGISUC2017 - New Lidar Technologies for 3DEP
NDGISUC2017 - New Lidar Technologies for 3DEPNDGISUC2017 - New Lidar Technologies for 3DEP
NDGISUC2017 - New Lidar Technologies for 3DEPNorth Dakota GIS Hub
 
Integrating GPS and SR Measures of Land in HH Surveys (Alberto Zezza, World B...
Integrating GPS and SR Measures of Land in HH Surveys (Alberto Zezza, World B...Integrating GPS and SR Measures of Land in HH Surveys (Alberto Zezza, World B...
Integrating GPS and SR Measures of Land in HH Surveys (Alberto Zezza, World B...ExternalEvents
 
PIAS 2013-GIS.pptxfskjczjsbchdbfscnnND dHSA
PIAS 2013-GIS.pptxfskjczjsbchdbfscnnND  dHSAPIAS 2013-GIS.pptxfskjczjsbchdbfscnnND  dHSA
PIAS 2013-GIS.pptxfskjczjsbchdbfscnnND dHSAFloridaTLaoaten
 
FOSS4G in Europe; Italy and the Politecnico de Milano
FOSS4G in Europe; Italy and the Politecnico de MilanoFOSS4G in Europe; Italy and the Politecnico de Milano
FOSS4G in Europe; Italy and the Politecnico de MilanoCarolina Arias Muñoz
 
Shallow Survey 2018 - Applications of Sonar Detection Uncertainty for Survey ...
Shallow Survey 2018 - Applications of Sonar Detection Uncertainty for Survey ...Shallow Survey 2018 - Applications of Sonar Detection Uncertainty for Survey ...
Shallow Survey 2018 - Applications of Sonar Detection Uncertainty for Survey ...Giuseppe Masetti
 
069MSW405_Devendra Tamrakar_Presentation
069MSW405_Devendra Tamrakar_Presentation069MSW405_Devendra Tamrakar_Presentation
069MSW405_Devendra Tamrakar_PresentationDevendra Tamrakar
 
Download-manuals-ground water-manual-gw-volume5operationmanualgiscreationofd...
 Download-manuals-ground water-manual-gw-volume5operationmanualgiscreationofd... Download-manuals-ground water-manual-gw-volume5operationmanualgiscreationofd...
Download-manuals-ground water-manual-gw-volume5operationmanualgiscreationofd...hydrologywebsite1
 
Download-manuals-ground water-manual-gw-volume5operationmanualgiscreationofd...
 Download-manuals-ground water-manual-gw-volume5operationmanualgiscreationofd... Download-manuals-ground water-manual-gw-volume5operationmanualgiscreationofd...
Download-manuals-ground water-manual-gw-volume5operationmanualgiscreationofd...hydrologyproject001
 
Grange, Stuart: Operation of the ICOS-Cities urban CO2 sensor network in Zuri...
Grange, Stuart: Operation of the ICOS-Cities urban CO2 sensor network in Zuri...Grange, Stuart: Operation of the ICOS-Cities urban CO2 sensor network in Zuri...
Grange, Stuart: Operation of the ICOS-Cities urban CO2 sensor network in Zuri...Integrated Carbon Observation System (ICOS)
 
Linked In Upload Gis
Linked In Upload   GisLinked In Upload   Gis
Linked In Upload Gisnadinwelsh
 
Do we need a new standard for visualizing the invisible?
Do we need a new standard for visualizing the invisible?Do we need a new standard for visualizing the invisible?
Do we need a new standard for visualizing the invisible?SANGHEE SHIN
 
Francisco J. Doblas-Big Data y cambio climático
Francisco J. Doblas-Big Data y cambio climáticoFrancisco J. Doblas-Big Data y cambio climático
Francisco J. Doblas-Big Data y cambio climáticoFundación Ramón Areces
 
TTI Production services
TTI Production servicesTTI Production services
TTI Production servicesTTI Production
 

Similar a Swiss Territorial Data Lab - geo Data Science - colloque FHNW (20)

LINUX Tag 2008: 4D Data Visualisation and Quality Control
LINUX Tag 2008: 4D Data Visualisation and Quality ControlLINUX Tag 2008: 4D Data Visualisation and Quality Control
LINUX Tag 2008: 4D Data Visualisation and Quality Control
 
New information sources for rain fields
New information sources for rain fieldsNew information sources for rain fields
New information sources for rain fields
 
BA Summit 2014 Predictive maintenance: Met big data het lek dichten
BA Summit 2014  Predictive maintenance: Met big data het lek dichtenBA Summit 2014  Predictive maintenance: Met big data het lek dichten
BA Summit 2014 Predictive maintenance: Met big data het lek dichten
 
Development of a soil carbon map for the United Republic of Tanzania
Development of a soil carbon map for the United Republic of TanzaniaDevelopment of a soil carbon map for the United Republic of Tanzania
Development of a soil carbon map for the United Republic of Tanzania
 
NDGISUC2017 - New Lidar Technologies for 3DEP
NDGISUC2017 - New Lidar Technologies for 3DEPNDGISUC2017 - New Lidar Technologies for 3DEP
NDGISUC2017 - New Lidar Technologies for 3DEP
 
Integrating GPS and SR Measures of Land in HH Surveys (Alberto Zezza, World B...
Integrating GPS and SR Measures of Land in HH Surveys (Alberto Zezza, World B...Integrating GPS and SR Measures of Land in HH Surveys (Alberto Zezza, World B...
Integrating GPS and SR Measures of Land in HH Surveys (Alberto Zezza, World B...
 
GeoCAPE Strategies
GeoCAPE StrategiesGeoCAPE Strategies
GeoCAPE Strategies
 
Lesson2 esa summer_school_brovelli
Lesson2 esa summer_school_brovelliLesson2 esa summer_school_brovelli
Lesson2 esa summer_school_brovelli
 
PIAS 2013-GIS.pptxfskjczjsbchdbfscnnND dHSA
PIAS 2013-GIS.pptxfskjczjsbchdbfscnnND  dHSAPIAS 2013-GIS.pptxfskjczjsbchdbfscnnND  dHSA
PIAS 2013-GIS.pptxfskjczjsbchdbfscnnND dHSA
 
FOSS4G in Europe; Italy and the Politecnico de Milano
FOSS4G in Europe; Italy and the Politecnico de MilanoFOSS4G in Europe; Italy and the Politecnico de Milano
FOSS4G in Europe; Italy and the Politecnico de Milano
 
T digest-update
T digest-updateT digest-update
T digest-update
 
Shallow Survey 2018 - Applications of Sonar Detection Uncertainty for Survey ...
Shallow Survey 2018 - Applications of Sonar Detection Uncertainty for Survey ...Shallow Survey 2018 - Applications of Sonar Detection Uncertainty for Survey ...
Shallow Survey 2018 - Applications of Sonar Detection Uncertainty for Survey ...
 
069MSW405_Devendra Tamrakar_Presentation
069MSW405_Devendra Tamrakar_Presentation069MSW405_Devendra Tamrakar_Presentation
069MSW405_Devendra Tamrakar_Presentation
 
Download-manuals-ground water-manual-gw-volume5operationmanualgiscreationofd...
 Download-manuals-ground water-manual-gw-volume5operationmanualgiscreationofd... Download-manuals-ground water-manual-gw-volume5operationmanualgiscreationofd...
Download-manuals-ground water-manual-gw-volume5operationmanualgiscreationofd...
 
Download-manuals-ground water-manual-gw-volume5operationmanualgiscreationofd...
 Download-manuals-ground water-manual-gw-volume5operationmanualgiscreationofd... Download-manuals-ground water-manual-gw-volume5operationmanualgiscreationofd...
Download-manuals-ground water-manual-gw-volume5operationmanualgiscreationofd...
 
Grange, Stuart: Operation of the ICOS-Cities urban CO2 sensor network in Zuri...
Grange, Stuart: Operation of the ICOS-Cities urban CO2 sensor network in Zuri...Grange, Stuart: Operation of the ICOS-Cities urban CO2 sensor network in Zuri...
Grange, Stuart: Operation of the ICOS-Cities urban CO2 sensor network in Zuri...
 
Linked In Upload Gis
Linked In Upload   GisLinked In Upload   Gis
Linked In Upload Gis
 
Do we need a new standard for visualizing the invisible?
Do we need a new standard for visualizing the invisible?Do we need a new standard for visualizing the invisible?
Do we need a new standard for visualizing the invisible?
 
Francisco J. Doblas-Big Data y cambio climático
Francisco J. Doblas-Big Data y cambio climáticoFrancisco J. Doblas-Big Data y cambio climático
Francisco J. Doblas-Big Data y cambio climático
 
TTI Production services
TTI Production servicesTTI Production services
TTI Production services
 

Más de Raphael Rollier

Swiss Territorial Data Lab - geo Data Science - Workshop CGC
Swiss Territorial Data Lab - geo Data Science - Workshop CGCSwiss Territorial Data Lab - geo Data Science - Workshop CGC
Swiss Territorial Data Lab - geo Data Science - Workshop CGCRaphael Rollier
 
Swiss Territorial Data Lab - geo Data Science - Forum SITG
Swiss Territorial Data Lab - geo Data Science - Forum SITG Swiss Territorial Data Lab - geo Data Science - Forum SITG
Swiss Territorial Data Lab - geo Data Science - Forum SITG Raphael Rollier
 
Swiss Territorial Data Lab - geo Data Science - colloque swisstopo
Swiss Territorial Data Lab - geo Data Science - colloque swisstopoSwiss Territorial Data Lab - geo Data Science - colloque swisstopo
Swiss Territorial Data Lab - geo Data Science - colloque swisstopoRaphael Rollier
 
Smart Tourist Destinations - leveraging Big Data & Blockchain
Smart Tourist Destinations - leveraging Big Data & BlockchainSmart Tourist Destinations - leveraging Big Data & Blockchain
Smart Tourist Destinations - leveraging Big Data & BlockchainRaphael Rollier
 
Connected Event - Cybersecurity 9 10 2018
Connected Event - Cybersecurity 9 10 2018Connected Event - Cybersecurity 9 10 2018
Connected Event - Cybersecurity 9 10 2018Raphael Rollier
 
Connected Event - Blockchain 3 05 2018
Connected Event - Blockchain 3 05 2018Connected Event - Blockchain 3 05 2018
Connected Event - Blockchain 3 05 2018Raphael Rollier
 
Ville durable, les facteurs de succès de la transition numérique
Ville durable, les facteurs de succès de la transition numériqueVille durable, les facteurs de succès de la transition numérique
Ville durable, les facteurs de succès de la transition numériqueRaphael Rollier
 
Start-up as innovation partner
Start-up as innovation partnerStart-up as innovation partner
Start-up as innovation partnerRaphael Rollier
 
Big Data for Big Impact - Swiss Data Day - EPFL - Nov 2017
Big Data for Big Impact - Swiss Data Day  - EPFL - Nov 2017Big Data for Big Impact - Swiss Data Day  - EPFL - Nov 2017
Big Data for Big Impact - Swiss Data Day - EPFL - Nov 2017Raphael Rollier
 
SmartCity workshop, Basel , April 2017
SmartCity workshop, Basel , April 2017SmartCity workshop, Basel , April 2017
SmartCity workshop, Basel , April 2017Raphael Rollier
 
Big Data in Real Estate - Digital Real Estate Summit
Big Data in Real Estate - Digital Real Estate SummitBig Data in Real Estate - Digital Real Estate Summit
Big Data in Real Estate - Digital Real Estate SummitRaphael Rollier
 
Smart Data Tool - Smart City Conference St. Gallen
Smart Data Tool - Smart City Conference St. GallenSmart Data Tool - Smart City Conference St. Gallen
Smart Data Tool - Smart City Conference St. GallenRaphael Rollier
 
Connected Event - Du Big Data au Smart Data 7Oct2015 - EPFL
Connected Event - Du Big Data au Smart Data 7Oct2015 - EPFLConnected Event - Du Big Data au Smart Data 7Oct2015 - EPFL
Connected Event - Du Big Data au Smart Data 7Oct2015 - EPFLRaphael Rollier
 
Smart City - Swisscom Dialogue Arena 18 juin2015, Lausanne
Smart City - Swisscom Dialogue Arena 18 juin2015, LausanneSmart City - Swisscom Dialogue Arena 18 juin2015, Lausanne
Smart City - Swisscom Dialogue Arena 18 juin2015, LausanneRaphael Rollier
 
Smart City Event EPFL 24 09 2014
Smart City Event EPFL 24 09 2014Smart City Event EPFL 24 09 2014
Smart City Event EPFL 24 09 2014Raphael Rollier
 

Más de Raphael Rollier (17)

Swiss Territorial Data Lab - geo Data Science - Workshop CGC
Swiss Territorial Data Lab - geo Data Science - Workshop CGCSwiss Territorial Data Lab - geo Data Science - Workshop CGC
Swiss Territorial Data Lab - geo Data Science - Workshop CGC
 
Swiss Territorial Data Lab - geo Data Science - Forum SITG
Swiss Territorial Data Lab - geo Data Science - Forum SITG Swiss Territorial Data Lab - geo Data Science - Forum SITG
Swiss Territorial Data Lab - geo Data Science - Forum SITG
 
Swiss Territorial Data Lab - geo Data Science - colloque swisstopo
Swiss Territorial Data Lab - geo Data Science - colloque swisstopoSwiss Territorial Data Lab - geo Data Science - colloque swisstopo
Swiss Territorial Data Lab - geo Data Science - colloque swisstopo
 
Smart Tourist Destinations - leveraging Big Data & Blockchain
Smart Tourist Destinations - leveraging Big Data & BlockchainSmart Tourist Destinations - leveraging Big Data & Blockchain
Smart Tourist Destinations - leveraging Big Data & Blockchain
 
Big Data, Big Profit ?
Big Data, Big Profit ?Big Data, Big Profit ?
Big Data, Big Profit ?
 
Connected Event - Cybersecurity 9 10 2018
Connected Event - Cybersecurity 9 10 2018Connected Event - Cybersecurity 9 10 2018
Connected Event - Cybersecurity 9 10 2018
 
Connected Event - Blockchain 3 05 2018
Connected Event - Blockchain 3 05 2018Connected Event - Blockchain 3 05 2018
Connected Event - Blockchain 3 05 2018
 
Ville durable, les facteurs de succès de la transition numérique
Ville durable, les facteurs de succès de la transition numériqueVille durable, les facteurs de succès de la transition numérique
Ville durable, les facteurs de succès de la transition numérique
 
Start-up as innovation partner
Start-up as innovation partnerStart-up as innovation partner
Start-up as innovation partner
 
Big Data for Big Impact - Swiss Data Day - EPFL - Nov 2017
Big Data for Big Impact - Swiss Data Day  - EPFL - Nov 2017Big Data for Big Impact - Swiss Data Day  - EPFL - Nov 2017
Big Data for Big Impact - Swiss Data Day - EPFL - Nov 2017
 
Mobility Flow in Zurich
Mobility Flow in ZurichMobility Flow in Zurich
Mobility Flow in Zurich
 
SmartCity workshop, Basel , April 2017
SmartCity workshop, Basel , April 2017SmartCity workshop, Basel , April 2017
SmartCity workshop, Basel , April 2017
 
Big Data in Real Estate - Digital Real Estate Summit
Big Data in Real Estate - Digital Real Estate SummitBig Data in Real Estate - Digital Real Estate Summit
Big Data in Real Estate - Digital Real Estate Summit
 
Smart Data Tool - Smart City Conference St. Gallen
Smart Data Tool - Smart City Conference St. GallenSmart Data Tool - Smart City Conference St. Gallen
Smart Data Tool - Smart City Conference St. Gallen
 
Connected Event - Du Big Data au Smart Data 7Oct2015 - EPFL
Connected Event - Du Big Data au Smart Data 7Oct2015 - EPFLConnected Event - Du Big Data au Smart Data 7Oct2015 - EPFL
Connected Event - Du Big Data au Smart Data 7Oct2015 - EPFL
 
Smart City - Swisscom Dialogue Arena 18 juin2015, Lausanne
Smart City - Swisscom Dialogue Arena 18 juin2015, LausanneSmart City - Swisscom Dialogue Arena 18 juin2015, Lausanne
Smart City - Swisscom Dialogue Arena 18 juin2015, Lausanne
 
Smart City Event EPFL 24 09 2014
Smart City Event EPFL 24 09 2014Smart City Event EPFL 24 09 2014
Smart City Event EPFL 24 09 2014
 

Último

DSPy a system for AI to Write Prompts and Do Fine Tuning
DSPy a system for AI to Write Prompts and Do Fine TuningDSPy a system for AI to Write Prompts and Do Fine Tuning
DSPy a system for AI to Write Prompts and Do Fine TuningLars Bell
 
Advanced Computer Architecture – An Introduction
Advanced Computer Architecture – An IntroductionAdvanced Computer Architecture – An Introduction
Advanced Computer Architecture – An IntroductionDilum Bandara
 
"Subclassing and Composition – A Pythonic Tour of Trade-Offs", Hynek Schlawack
"Subclassing and Composition – A Pythonic Tour of Trade-Offs", Hynek Schlawack"Subclassing and Composition – A Pythonic Tour of Trade-Offs", Hynek Schlawack
"Subclassing and Composition – A Pythonic Tour of Trade-Offs", Hynek SchlawackFwdays
 
TeamStation AI System Report LATAM IT Salaries 2024
TeamStation AI System Report LATAM IT Salaries 2024TeamStation AI System Report LATAM IT Salaries 2024
TeamStation AI System Report LATAM IT Salaries 2024Lonnie McRorey
 
New from BookNet Canada for 2024: BNC CataList - Tech Forum 2024
New from BookNet Canada for 2024: BNC CataList - Tech Forum 2024New from BookNet Canada for 2024: BNC CataList - Tech Forum 2024
New from BookNet Canada for 2024: BNC CataList - Tech Forum 2024BookNet Canada
 
Transcript: New from BookNet Canada for 2024: Loan Stars - Tech Forum 2024
Transcript: New from BookNet Canada for 2024: Loan Stars - Tech Forum 2024Transcript: New from BookNet Canada for 2024: Loan Stars - Tech Forum 2024
Transcript: New from BookNet Canada for 2024: Loan Stars - Tech Forum 2024BookNet Canada
 
SALESFORCE EDUCATION CLOUD | FEXLE SERVICES
SALESFORCE EDUCATION CLOUD | FEXLE SERVICESSALESFORCE EDUCATION CLOUD | FEXLE SERVICES
SALESFORCE EDUCATION CLOUD | FEXLE SERVICESmohitsingh558521
 
Sample pptx for embedding into website for demo
Sample pptx for embedding into website for demoSample pptx for embedding into website for demo
Sample pptx for embedding into website for demoHarshalMandlekar2
 
The Role of FIDO in a Cyber Secure Netherlands: FIDO Paris Seminar.pptx
The Role of FIDO in a Cyber Secure Netherlands: FIDO Paris Seminar.pptxThe Role of FIDO in a Cyber Secure Netherlands: FIDO Paris Seminar.pptx
The Role of FIDO in a Cyber Secure Netherlands: FIDO Paris Seminar.pptxLoriGlavin3
 
Take control of your SAP testing with UiPath Test Suite
Take control of your SAP testing with UiPath Test SuiteTake control of your SAP testing with UiPath Test Suite
Take control of your SAP testing with UiPath Test SuiteDianaGray10
 
unit 4 immunoblotting technique complete.pptx
unit 4 immunoblotting technique complete.pptxunit 4 immunoblotting technique complete.pptx
unit 4 immunoblotting technique complete.pptxBkGupta21
 
How to write a Business Continuity Plan
How to write a Business Continuity PlanHow to write a Business Continuity Plan
How to write a Business Continuity PlanDatabarracks
 
Dev Dives: Streamline document processing with UiPath Studio Web
Dev Dives: Streamline document processing with UiPath Studio WebDev Dives: Streamline document processing with UiPath Studio Web
Dev Dives: Streamline document processing with UiPath Studio WebUiPathCommunity
 
Transcript: New from BookNet Canada for 2024: BNC CataList - Tech Forum 2024
Transcript: New from BookNet Canada for 2024: BNC CataList - Tech Forum 2024Transcript: New from BookNet Canada for 2024: BNC CataList - Tech Forum 2024
Transcript: New from BookNet Canada for 2024: BNC CataList - Tech Forum 2024BookNet Canada
 
Moving Beyond Passwords: FIDO Paris Seminar.pdf
Moving Beyond Passwords: FIDO Paris Seminar.pdfMoving Beyond Passwords: FIDO Paris Seminar.pdf
Moving Beyond Passwords: FIDO Paris Seminar.pdfLoriGlavin3
 
Training state-of-the-art general text embedding
Training state-of-the-art general text embeddingTraining state-of-the-art general text embedding
Training state-of-the-art general text embeddingZilliz
 
What is Artificial Intelligence?????????
What is Artificial Intelligence?????????What is Artificial Intelligence?????????
What is Artificial Intelligence?????????blackmambaettijean
 
Artificial intelligence in cctv survelliance.pptx
Artificial intelligence in cctv survelliance.pptxArtificial intelligence in cctv survelliance.pptx
Artificial intelligence in cctv survelliance.pptxhariprasad279825
 
DevoxxFR 2024 Reproducible Builds with Apache Maven
DevoxxFR 2024 Reproducible Builds with Apache MavenDevoxxFR 2024 Reproducible Builds with Apache Maven
DevoxxFR 2024 Reproducible Builds with Apache MavenHervé Boutemy
 
SIP trunking in Janus @ Kamailio World 2024
SIP trunking in Janus @ Kamailio World 2024SIP trunking in Janus @ Kamailio World 2024
SIP trunking in Janus @ Kamailio World 2024Lorenzo Miniero
 

Último (20)

DSPy a system for AI to Write Prompts and Do Fine Tuning
DSPy a system for AI to Write Prompts and Do Fine TuningDSPy a system for AI to Write Prompts and Do Fine Tuning
DSPy a system for AI to Write Prompts and Do Fine Tuning
 
Advanced Computer Architecture – An Introduction
Advanced Computer Architecture – An IntroductionAdvanced Computer Architecture – An Introduction
Advanced Computer Architecture – An Introduction
 
"Subclassing and Composition – A Pythonic Tour of Trade-Offs", Hynek Schlawack
"Subclassing and Composition – A Pythonic Tour of Trade-Offs", Hynek Schlawack"Subclassing and Composition – A Pythonic Tour of Trade-Offs", Hynek Schlawack
"Subclassing and Composition – A Pythonic Tour of Trade-Offs", Hynek Schlawack
 
TeamStation AI System Report LATAM IT Salaries 2024
TeamStation AI System Report LATAM IT Salaries 2024TeamStation AI System Report LATAM IT Salaries 2024
TeamStation AI System Report LATAM IT Salaries 2024
 
New from BookNet Canada for 2024: BNC CataList - Tech Forum 2024
New from BookNet Canada for 2024: BNC CataList - Tech Forum 2024New from BookNet Canada for 2024: BNC CataList - Tech Forum 2024
New from BookNet Canada for 2024: BNC CataList - Tech Forum 2024
 
Transcript: New from BookNet Canada for 2024: Loan Stars - Tech Forum 2024
Transcript: New from BookNet Canada for 2024: Loan Stars - Tech Forum 2024Transcript: New from BookNet Canada for 2024: Loan Stars - Tech Forum 2024
Transcript: New from BookNet Canada for 2024: Loan Stars - Tech Forum 2024
 
SALESFORCE EDUCATION CLOUD | FEXLE SERVICES
SALESFORCE EDUCATION CLOUD | FEXLE SERVICESSALESFORCE EDUCATION CLOUD | FEXLE SERVICES
SALESFORCE EDUCATION CLOUD | FEXLE SERVICES
 
Sample pptx for embedding into website for demo
Sample pptx for embedding into website for demoSample pptx for embedding into website for demo
Sample pptx for embedding into website for demo
 
The Role of FIDO in a Cyber Secure Netherlands: FIDO Paris Seminar.pptx
The Role of FIDO in a Cyber Secure Netherlands: FIDO Paris Seminar.pptxThe Role of FIDO in a Cyber Secure Netherlands: FIDO Paris Seminar.pptx
The Role of FIDO in a Cyber Secure Netherlands: FIDO Paris Seminar.pptx
 
Take control of your SAP testing with UiPath Test Suite
Take control of your SAP testing with UiPath Test SuiteTake control of your SAP testing with UiPath Test Suite
Take control of your SAP testing with UiPath Test Suite
 
unit 4 immunoblotting technique complete.pptx
unit 4 immunoblotting technique complete.pptxunit 4 immunoblotting technique complete.pptx
unit 4 immunoblotting technique complete.pptx
 
How to write a Business Continuity Plan
How to write a Business Continuity PlanHow to write a Business Continuity Plan
How to write a Business Continuity Plan
 
Dev Dives: Streamline document processing with UiPath Studio Web
Dev Dives: Streamline document processing with UiPath Studio WebDev Dives: Streamline document processing with UiPath Studio Web
Dev Dives: Streamline document processing with UiPath Studio Web
 
Transcript: New from BookNet Canada for 2024: BNC CataList - Tech Forum 2024
Transcript: New from BookNet Canada for 2024: BNC CataList - Tech Forum 2024Transcript: New from BookNet Canada for 2024: BNC CataList - Tech Forum 2024
Transcript: New from BookNet Canada for 2024: BNC CataList - Tech Forum 2024
 
Moving Beyond Passwords: FIDO Paris Seminar.pdf
Moving Beyond Passwords: FIDO Paris Seminar.pdfMoving Beyond Passwords: FIDO Paris Seminar.pdf
Moving Beyond Passwords: FIDO Paris Seminar.pdf
 
Training state-of-the-art general text embedding
Training state-of-the-art general text embeddingTraining state-of-the-art general text embedding
Training state-of-the-art general text embedding
 
What is Artificial Intelligence?????????
What is Artificial Intelligence?????????What is Artificial Intelligence?????????
What is Artificial Intelligence?????????
 
Artificial intelligence in cctv survelliance.pptx
Artificial intelligence in cctv survelliance.pptxArtificial intelligence in cctv survelliance.pptx
Artificial intelligence in cctv survelliance.pptx
 
DevoxxFR 2024 Reproducible Builds with Apache Maven
DevoxxFR 2024 Reproducible Builds with Apache MavenDevoxxFR 2024 Reproducible Builds with Apache Maven
DevoxxFR 2024 Reproducible Builds with Apache Maven
 
SIP trunking in Janus @ Kamailio World 2024
SIP trunking in Janus @ Kamailio World 2024SIP trunking in Janus @ Kamailio World 2024
SIP trunking in Janus @ Kamailio World 2024
 

Swiss Territorial Data Lab - geo Data Science - colloque FHNW

  • 1. FHNW COLLOQUIUM March 16, 2021 SWISS TERRITORIAL DATA LAB APPLIED DATA SCIENCE Raphael Rollier - Nils Hamel – Huriel Reichel Adrian Meyer - Christian Dettwiler
  • 2.  Introduction about the Swiss Territorial Data Lab  4D Platform  Register of Buildings and Dwellings  Objects Detection  A Cantonal perspective  Q&A Program
  • 3. A co-creation project A way to share and replicate A space for experimentation Swiss Territorial Data Lab (STDL) is...
  • 4. If you want to go fast, go alone. If you want to go far, go together STDL partners are…
  • 5. Solving concrete problems in public administrations with Geo Data Science STDL mission is…
  • 6. What is the most effective way to find out the construction date of buildings to complete the Building Register ? How can I detect automatically changes in the field in order to update the Land Register more rapidly ? How can I improve the monitoring of solar energy usage by detecting automatically panel installations ? How can I monitor more effectively the development of mining ? The type of challenges we are exploring…
  • 7. FHNW COLLOQUIUM March 16, 2021 THE TIME DIMENSION RBD COMPLETION RESEARCH PROJECT Nils Hamel – Huriel Reichel
  • 8. STDL 4D PLATFORM SRTM EXAMPLE – 470 GB (ASC)
  • 9. 2009-10 2013-04 2017-04 THE TIME DIMENSION EXAMPLE OF GENEVA LIDAR – 250 GB (LAS)
  • 10. THURGAU – 2020-10-17 INTERLIS – ITF THURGAU – 2020-10-13 INTERLIS – ITF
  • 12. REGISTER OF BUILDINGS & DWELLINGS Federal Statistical Office (OFS/BFS) & STDL ● Federal Register ● Missing buildings construction years ● Automating construction years gathering Two complementary research approaches PROJECT
  • 13. NATIONAL MAPS Using swiss 1:25’000 national maps ● Tracking the buildings on the maps ● Detection of their appearence ● Covering 2020 to 1950 APPROACH
  • 14. SWISS NATIONAL MAPS HOMOGENEOUS AND STABLE METHODOLOGY Emergence of the notion of 3D raster 2010 2004 1998 1993
  • 15.
  • 16.
  • 17. DEDUCTION PROCESS DETECTIONS & MORPHOLOGICAL CRITERION Simple case Complex case
  • 18. VALIDATION METRIC RESULTS ASSESSMENT Manually gathered sets of synchronous buildings Register : 1962 – National Maps : 1960-1964
  • 19. RESULTS ASSESSMENT NATIONAL MAPS APPROACH With an average temporal resolution of 5.8 years : 84.7%
  • 20.
  • 21. STATISTICAL Using statistical urban model ● Workaround the lack of maps ● Improve construction years approximation ● Covering years before 1950 APPROACH
  • 22. THEORETICAL BACKGROUND BURGESS URBAN MODEL Pattern in urban growth & spatial dependence
  • 23. METHODOLOGY FILLING THE GAPS IN THE DATABASE Temporal variance to compute spatial radii
  • 26. RESULTS ASSESSMENT STATISTICAL APPROACH 95% of building correctly placed within a 31 years interval
  • 27. ● National Maps Approach 84.7% within ±5.8 years ● Statistical Approach 95 % within ±31 years ● The importance of Time Dimension Provides relevant information CONCLUSION
  • 28. FHNW COLLOQUIUM March 16, 2021 OBJECT DETECTION FRAMEWORK Adrian Meyer
  • 29. STDL OBJECT DETECTION FRAMEWORK Generating a Model from Cadastral Vectors and Aerial Images Predicting Objects in the Same or a New Area of Interest TRAINING with Known Objects INFERENCE for Unknown Objects BASIC IDEA
  • 30. WORKFLOW TRANSFER LEARNING Massive Training Dataset A Acquired Knowledge Deep Learning System B(A) Small Specific Dataset B Exported from Cadastre Deep Learning System A New Predictions Source: https://cdn-media-1.freecodecamp.org/images/1*lMEd6AcDmpH0mDzBHyiERw.png Source: https://medium.com/swlh/object-detection-and-instance-segmentation-a-detailed-overview-94ca109274f2
  • 31. Area of Interest AoI Training AoI with Pool Labels Prediction AoI
  • 32. ZONING PLAN Canton Thurgau LAYER WATER BASINS Cadastre Export Exclusion of Industrial Areas
  • 33. AoI Tiling Legend Cantonal Boundary AoI Boundary Labels to be checked Outlier Labels (discarded)
  • 34. TRAINING Ground Truth Generation: Dataset Evaluation Split 80% Training Used to Train Model Weights 10% Validation Tuning Model Parameters 10% Test Unbiased Assessment Ground Truth Labels
  • 36. RESULTS How Good Did We Do? # True Positives # False Negatives # False Positives
  • 37. Registered & Undetected (FN) Detected, but not registered (FP) Registered & Detected (TP) Wikimedia Commons, 2021
  • 38. F1 Score The F1 Score is the Harmonic Mean of Precision and Recall. Wikimedia Commons (2021) P. Mirla (2018), via github.io
  • 39. Metrics Test Dataset at Zoom Level 19 Trained on SWISSIMAGE from Geneva and Neuchatel Precision and Recall are a Function of the Confidence 84%
  • 40. Thurgau Predictions: Cadastre Updates True Positives • Threshold THR ≥ 5% • 2‘227 of 2‘959 Pools detected • 75% Detection Success  25% (732) «wrongly» in Cadastre ?
  • 41. Thurgau Predictions: Cadastre Updates False Negatives • 732 of 2‘959 Segments >5m² missed  25% «wrongly» in cadastre?
  • 42. Thurgau Predictions: Cadastre Updates False Positives • Threshold THR ≥ 97% • 271 non-listed Pools  9% missing in Cadastre? • Threshold THR ≥ 94% • 672 non-listed Pools  23% missing in Cadastre?
  • 43. Manual Evaluation Frauenfeld Visible Pools Dataset Metrics True Positive: 81 False Positive: 9 False Negative: 18 F1-Score: 85.7% Precision: 90.0% Recall: 81.8% - - - - - - - - - - - - Detector Metrics True Positive: 94 False Positive: 16 False Negative: 5 F1-Score: 90.0% Precision: 85.5% Recall: 94.9%
  • 44. Swimming Pools: Zoom Level Results Zoom Level 15 ≈ 480 cm/px GSD 16 ≈ 240 cm/px GSD 17 ≈ 120 cm/px GSD 18 ≈ 60 cm/px GSD 19* ≈ 30 cm/px GSD File System Load 1’949 Tiles = 0.4 GB 4’216 Tiles = 1.3 GB 12’817 Tiles = 4.0 GB 42’990 Tiles = 11 GB 154’861 Tiles = 40 GB Processing Duration (Prep. / DL+Pred. / Postp.) ± 25 min ± 40 min ± 120 min = 2 h ± 240 min = 4 h ± 1’200 min = 20 h Max. F1 Score on TST Dataset 55.5 % 75.3 % 82.5 % 83.4 % 84.1 % * Zoom Level 20 contains 583’014 Tiles = 179 GB and was expected to take ~120h, but aborted prematurely due to bandwidth memory errors
  • 45. What’s Next for the End Users? Update Cadastre Send Letters to Swimming Pool Owners Think about Detectable Objects and Contact Us!
  • 46. Silage Bale Detector Labeling Strategy - Manually Digitizing 200 Stacks of Silage Bales in QGIS - Training a Preliminary Detector - Use 300 Highest-Confidence Predictions as New Labels - Manual Correction and Filling In of Complete Training Tiles Result 700 Labels in 1.5 Days Wikimedia Commons (2021)
  • 47. New Elements: Silage Bales Zoom Level 16 ≈ 240 cm/px GSD 17 ≈ 120 cm/px GSD 18 ≈ 60 cm/px GSD 19 ≈ 30 cm/px GSD 20 ≈ 15 cm/px GSD File System Load 8’000 Tiles = 1.3 GB 25’000 Tiles = 8.0 GB 84’000 Tiles = 26 GB 310’000 Tiles = 80 GB 1’310’000 Tiles = 320 GB Processing Duration (Prep. / DL+Pred. / Postp.) ± 40 min ± 120 min = 2 h ± 240 min = 4 h ± 900 min = 15 h ± 6’000 min = 100 h Max. F1 Score on TST Dataset 52.5 % 74.7 % 87.2 % 92.3 % 90.9 %
  • 50. SWISS TERRITORIAL DATA LAB APPLIED DATA SCIENCE Sicht eines Kantons: Thurgau (Einbezogen in die Projekte 4D, Schwimmbäder, Ko- Produktion Hoheitsgrenzen, Siloballen)
  • 51. A Cantonal perspective Potenziale aus Sicht eines Kantons: 4D-Plattform  Die amtliche Vermessung ist Referenzdatensatz! Änderungen wirken sich auf viele «aufbauende Geodaten» aus. => Information wertvoll  Datenmigration auf neue IT, neues Datenmodell: Bleiben die Daten korrekt und vollständig?  Bauverwaltungen: Entsprechen die Änderungen den neu bewilligten Bauten? Gibt es nicht bewilligte Bauten? SECTION
  • 52. A Cantonal perspective Potenziale aus Sicht eines Kantons: Objektdetektion  Schwimmbäder: Im TG eher sekundär, als Test der KI-Tools wertvoll  Es gibt viele andere Potenziale: Bsp: Siloballen, Echo des Landwirtschaftsamtes: Begeisterung Die Objektdetektion sollte nicht auf das Thema amtliche Vermessung eingeschränkt werden. Die Kunden haben vielfältige Interessen, diese sollten wir «abholen». Das stärkt unsere Position. SECTION
  • 53. A Cantonal perspective Potenziale aus Sicht eines Kantons: STDL = Toolbox  Die STDL-Tools können mit etwas Phantasie und vor allem mit Kenntnis der Kundenbedürfnisse breit eingesetzt werden. Leitsatz Amt für Geoinformation Thurgau: Wir schaffen mit Geoinformation volkswirtschaft- lichen Nutzen. Die STDL-Toolbox hilft uns dabei. SECTION
  • 54. FHNW COLLOQUIUM March 16, 2021 SWISS TERRITORIAL DATA LAB APPLIED DATA SCIENCE Nils Hamel – Huriel Reichel – Adrian Meyer Raphael Rollier – Christian Dettwiler info@stdl.ch stdl.ch