Roelof Pieters (Overstory) – Tackling Forest Fires and Deforestation with Satellite Data and AI

C
Roelof Pieters
CTO & Co-founder
roelof@overstory.ai
@graphific
Tackling Forest Fires and Deforestation with Satellite
Data and AI
May 14, 2021
• Computer Science/Deep Learning
• Anthropology and Development
Sociology
• Previously founded Creative.ai,
Stockholm.ai, GitXiv and other open
source initiatives/networks/startups
Introduction
Roelof Pieters
CTO & Co-founder, Overstory
This session
• Advancements in satellite data & AI
• Example: deforestation monitoring
• Example: preventing wildfires and power
outages
Indra den Bakker
CEO & Co-founder
Anniek Schouten
COO & Co-founder
David de Meij
Data Scientist
Roelof Pieters
CTO & Co-founder
Rochelle Silva
Radar &
GIS Researcher
Lorenzo Riches
Data Scientist
We’re Overstory
Arsha Yuditha
Amiranti
GIS Specialist
Elvira Garkava
Business Developer
Fiona Spruill
Chief Product Officer
Andrea Giardini
DevOps Engineer
Killian Tobin
Head of Business
Development
Customers
Investors
Partners
The problem
Insect outbreaks
Forest fires
Deforestation
Forest fire (Source: Nasa)
Bark beetle infestation (Source: Sodra)
Deforestation (Source: ESA, Google)
Remote Sensing: an overview
Satellite market
Today 2029
# of satellites
2010
1000
800
600
400
200
1,470 satellites launched
between 2010 and 2019
8,500 satellites expected to launch
between 2019 and 2028*
Hyperspectral
VHR SAR Video Real-time
Source:Euroconsult: https://spacenews.com/analysis-are-smallsats-entering-the-maturity-stage/
Input data High-resolution satellite imagery
up to 30 cm resolution
Input data Invisible light like near-IR
Multispectral data
Input data Weather independent radar images
SAR data
July 2, 2017 July 4, 2017 July 5, 2017
Input data Frequent revisits
up to daily on a global scale
Input data Frequent revisits
up to daily on a global scale
2.5m
source:
https://platform.digitalglobe.com/earth-imaging-basics-spatial-resolution/
30cm
Tri-stereo and video monitoring
for 3D-mapping
SAR data to look through
clouds
Multi- and hyperspectral
data
Up to 30 cm resolution
Up to 20 VHR daily revisits
& geostationary satellites
Very high-resolution satellite data
● Mega Large Imagery
● Constant change
● Noisy satellite data
Data Science Challenges
● Noisy or lack of labels
● Generalize from training
● “ground truth”
● Classical machine learning still the norm
Unsupervised Machine Learning
???
A? B? C?
Allows Overstory to get insights
anywhere in the world with 1-2 factors
less customer/labeled data (1000 instead
of 10K-100K data points)
"Everything is related to everything else,
but near things are more related than
distant things"
Stereo Imaging
For creating height maps and 3D maps we use Stereo Imaging
techniques to create Digital Surface Maps (DSM) and Digital Terrain
Maps (DTM), as well as 3D point clouds as in this example
Stereo Imaging Learning Feature Matching with Graph Neural Networks
Magic Leap/ETH Zurich (2020)
A process by which freely accessible low resolution imagery can be upscaled to commercial-grade high
resolution, allowing for more accurate insights, easier labelling by our annotators, and increased
accuracy for our machine learning models, at a cheaper cost
(Upscaling Landsat-8 to
Digital Globe Worldview-3-like level)
LS8 TIRS
LS8
CIRRUS
LS8
Panchromatic
LS8 SWIR
LS8 NIR LS8 RGB
DG-WV3 RGB
DG-WV3
Panchromatic
DG-WV3 SWIR
31cm
1.24m
3.7m
15m
100m
30m
Generative Upscaling
Li et al (2019) Feedback Network for Image
Super-Resolution"
See also Super-Resolution Generative
Adversarial Network(s) (many papers)
Active Learning
(re)train
candidate selection
oracle / human annotator
● eg BAyesian Active Learning library
(BaaL) by ElementAI:
https://github.com/ElementAI/baal/
○ MCDropout (Gal et al. 2015)
○ BALD (Houlsby et al. 2011)
(ElementAI)
Living data repository
training data for scalable and
accurate deep learning
algorithms
case #1: Machine Learning for
deforestation monitoring
● high resolution forest and landcover map
● up to date with 2014 to now
● high forest and crop type
accuracy
Deforestation monitoring
what how
● deep learning for segmentation
● multiple satellite sources: sensor
fusion / multimodality
● noisy data: generative gap filling
● dynamic data regime: open data, active
labelling (ground and satellite), noisy
labels
Climate &
weather
Height data
Advancements in
deep learning
SAR (radar)
data
Multi resolution
data
Sensor fusion
Convolutional Neural
Networks
Unsupervised
learning and active
(bayesian) learning
Multispectral
data
Output
Segmentation
Layers of satellite
imagery
Land Cover Segmentation on
pixel level
Active Learning
Encoder Neural Planet Embedding
1.
2.
3.
4a.
Data
repositor
y
5.
4b.
(Overstory training data
pipeline)
Annotator
Field worker
Public Data
Roelof Pieters (Overstory) – Tackling Forest Fires and Deforestation with Satellite Data and AI
Roelof Pieters (Overstory) – Tackling Forest Fires and Deforestation with Satellite Data and AI
High-Resolution Global Maps of 21st-
Century Forest Cover Change
(Hansen et al., Science 2013).
Overstory 10m and less
Hansen et al. 2013/2019 1.7 (2000)
GlobCover 2010 2.3 (2009)
CCI Landcover Africa 20m (2016)
case # 2: Scaling ML for very high
resolution earth observation & risk
monitoring
Vegetation &
powerlines
of global annual
CO2 emissions
~1%
direct
economic losses
$21B
people without
power
40M
Roelof Pieters (Overstory) – Tackling Forest Fires and Deforestation with Satellite Data and AI
Roelof Pieters (Overstory) – Tackling Forest Fires and Deforestation with Satellite Data and AI
● very high resolution tree species map:
37 different tree species, shrub species and
grass classes
● over 16,000 km of power lines
with a corridor width of 150
meter
● total area of over 505,000 km2
● hard constraints on minimum
level of accuracy
fire risk monitoring
what how
● deep learning, naturally :)
● open source labels, customer labels,
external party ground validation
● very high resolution satellite imagery
(50cm)
● massively parallel distributed
processing through dask, kubernetes,
and distributed data parallel training
Jupyterhub
Jupyterhub allows users to create
dedicated computational environments
Our Infrastructure
Getting a new notebook
Our Infrastructure - Jupyterhub
👥
I need a new notebook! Processing...
We need a new machine
for this...
New node
Notebook available
1
2
3
4
5
6
Dask
Dask provides advanced parallelism for
analytics, enabling performance at scale
for the tools you love
Our Infrastructure
I need some heavy compute
Our Infrastructure - Dask
👥
I need a new
Dask cluster! Dask-gateway
Dask-scheduler
Dask-worker
Dask-worker
Cluster
available
1
2
3
4
5
6
Dask cluster - Structure
Our Infrastructure - Dask
Dask-gateway
Dask-scheduler
Dask-scheduler
Dask-worker Dask-worker
Dask-worker Dask-worker
Dask-worker Dask-worker
Dask-worker Dask-worker
Dask-worker Dask-worker
Dask-worker Dask-worker
Papermill
Papermill provides queues and ETL
capabilities for running and logging jupyter
notebook workloads
Our Infrastructure
Our Infrastructure - Papermill
Image by Matthew Seal,
Netflix
Our Infrastructure - Papermill
Image by Matthew
Seal, Netflix
All together now
Our Infrastructure
Abernathey et al. (2021):
Pangeo architecture
diagram.
Jupyterhub
Dask Dashboard
Monitoring Kubernetes/Dask with Grafana
Real-time
Vegetation
Intelligence
Platform
Tree height
Canopy segmentation
Fire risks
Tree species
Verification of trimming
Storm damage
Risk identification
Predictive prioritization
Overstory is on a mission to monitor all
natural resources on Earth in real-time.
We’re in this together!
Overstory is on a mission to monitor all
natural resources on Earth in real-time.
Come join us!
https://www.overstory.com/careers
Learning Resources / Research
● List of videos about Geospatial data science from FOSS4G (Free
and Open Source Software for Geospatial / OSGeo)
https://www.youtube.com/channel/UC_2Lyc9VUX-jC-E1prJitHw/vi
deos
● ICLR 2020 proceedings now available: https://iclr.cc/virtual_2020/
& video recordings of climate change AI workshop:
https://www.youtube.com/channel/UCyjDr_aoMlzhSvCTdT7eZ9g/
videos
● CVPR pre-papers for EarthVision: Large Scale Computer Vision
for Remote Sensing Imagery Workshop
http://openaccess.thecvf.com/CVPR2020_workshops/CVPR2020
_w11.py
Awesome lists of resources:
● https://github.com/sshuair/awesome-gis
● https://github.com/robmarkcole/satellite-image-deep-learning
● https://github.com/chrieke/awesome-satellite-imagery-datasets
● https://github.com/acgeospatial/awesome-earthobservation-code
● https://github.com/wenhwu/awesome-remote-sensing-change-
detection
Amazing projects
● https://www.globalforestwatch.org/
● https://trase.earth/
● https://www.half-earthproject.org/
● https://www.climatewatchdata.org/
Geospatial Toolkit/UI:
● QGIS: https://www.qgis.org
Satellite Data
● sentinel-2 10m resolution satellite imagery:
https://scihub.copernicus.eu/dhus/#/home
● landsat 30m resolution satellite imagery:
https://landsat.gsfc.nasa.gov/
ML for geospatial/satellite data libraries:
● https://github.com/sentinel-hub/eo-learn
● https://rastervision.io/
● https://github.com/fastai/fastai2/
Resources
Rolnick, et al. Tackling Climate Change with Machine Learning, arXiv:1906.05433 & https://www.climatechange.ai/
Roelof Pieters (Overstory) – Tackling Forest Fires and Deforestation with Satellite Data and AI
www.overstory.ai
linkedin.com/company/overstoryai
twitter.com/overstoryai
medium.com/overstoryai
Roelof Pieters
CTO & Co-founder
roelof@overstory.ai
@graphific
1 de 52

Recomendados

Low level feature extraction - chapter 4 por
Low level feature extraction - chapter 4Low level feature extraction - chapter 4
Low level feature extraction - chapter 4Aalaa Khattab
1.2K visualizações52 slides
Spatial enhancement por
Spatial enhancement Spatial enhancement
Spatial enhancement abinarkt
4.2K visualizações25 slides
Remote Sensing Imagery & Artificial Intelligence por
Remote Sensing Imagery & Artificial IntelligenceRemote Sensing Imagery & Artificial Intelligence
Remote Sensing Imagery & Artificial IntelligenceEsri Ireland
1.9K visualizações19 slides
Image quality assessment and statistical evaluation por
Image quality assessment and statistical evaluationImage quality assessment and statistical evaluation
Image quality assessment and statistical evaluationDocumentStory
2.3K visualizações63 slides
SDSC Technology Forum: Increasing the Impact of High Resolution Topography Da... por
SDSC Technology Forum: Increasing the Impact of High Resolution Topography Da...SDSC Technology Forum: Increasing the Impact of High Resolution Topography Da...
SDSC Technology Forum: Increasing the Impact of High Resolution Topography Da...OpenTopography Facility
90 visualizações34 slides
SPAR 2015 - Civil Maps Presentation by Sravan Puttagunta por
SPAR 2015 - Civil Maps Presentation by Sravan PuttaguntaSPAR 2015 - Civil Maps Presentation by Sravan Puttagunta
SPAR 2015 - Civil Maps Presentation by Sravan PuttaguntaSravan Puttagunta
1.3K visualizações23 slides

Mais conteúdo relacionado

Similar a Roelof Pieters (Overstory) – Tackling Forest Fires and Deforestation with Satellite Data and AI

FME UC 2014: Hexagon Keynote por
FME UC 2014: Hexagon KeynoteFME UC 2014: Hexagon Keynote
FME UC 2014: Hexagon KeynoteSafe Software
1.9K visualizações82 slides
Irish Earth Observation Symposium 2014: EO Technologies - The Cause and Solut... por
Irish Earth Observation Symposium 2014: EO Technologies - The Cause and Solut...Irish Earth Observation Symposium 2014: EO Technologies - The Cause and Solut...
Irish Earth Observation Symposium 2014: EO Technologies - The Cause and Solut...IMGS
854 visualizações37 slides
Q4 2016 GeoTrellis Presentation por
Q4 2016 GeoTrellis PresentationQ4 2016 GeoTrellis Presentation
Q4 2016 GeoTrellis PresentationRob Emanuele
963 visualizações176 slides
understanding the planet using satellites and deep learning por
understanding the planet using satellites and deep learningunderstanding the planet using satellites and deep learning
understanding the planet using satellites and deep learningAlbert Pujol Torras
64 visualizações22 slides
IMGS Geospatial User Group 2014 - Big data management with Apollo por
IMGS Geospatial User Group 2014 - Big data management with ApolloIMGS Geospatial User Group 2014 - Big data management with Apollo
IMGS Geospatial User Group 2014 - Big data management with ApolloIMGS
365 visualizações25 slides
Drones and A.I in Earth Science por
Drones and A.I in Earth ScienceDrones and A.I in Earth Science
Drones and A.I in Earth ScienceARDC
969 visualizações15 slides

Similar a Roelof Pieters (Overstory) – Tackling Forest Fires and Deforestation with Satellite Data and AI(20)

FME UC 2014: Hexagon Keynote por Safe Software
FME UC 2014: Hexagon KeynoteFME UC 2014: Hexagon Keynote
FME UC 2014: Hexagon Keynote
Safe Software1.9K visualizações
Irish Earth Observation Symposium 2014: EO Technologies - The Cause and Solut... por IMGS
Irish Earth Observation Symposium 2014: EO Technologies - The Cause and Solut...Irish Earth Observation Symposium 2014: EO Technologies - The Cause and Solut...
Irish Earth Observation Symposium 2014: EO Technologies - The Cause and Solut...
IMGS854 visualizações
Q4 2016 GeoTrellis Presentation por Rob Emanuele
Q4 2016 GeoTrellis PresentationQ4 2016 GeoTrellis Presentation
Q4 2016 GeoTrellis Presentation
Rob Emanuele963 visualizações
understanding the planet using satellites and deep learning por Albert Pujol Torras
understanding the planet using satellites and deep learningunderstanding the planet using satellites and deep learning
understanding the planet using satellites and deep learning
Albert Pujol Torras64 visualizações
IMGS Geospatial User Group 2014 - Big data management with Apollo por IMGS
IMGS Geospatial User Group 2014 - Big data management with ApolloIMGS Geospatial User Group 2014 - Big data management with Apollo
IMGS Geospatial User Group 2014 - Big data management with Apollo
IMGS365 visualizações
Drones and A.I in Earth Science por ARDC
Drones and A.I in Earth ScienceDrones and A.I in Earth Science
Drones and A.I in Earth Science
ARDC969 visualizações
Planet OS: Indexing the Real World (a lecture at the Stanford Engineering Sch... por Rainer Sternfeld
Planet OS: Indexing the Real World (a lecture at the Stanford Engineering Sch...Planet OS: Indexing the Real World (a lecture at the Stanford Engineering Sch...
Planet OS: Indexing the Real World (a lecture at the Stanford Engineering Sch...
Rainer Sternfeld631 visualizações
Rainer Sternfeld - Planetary Big Data - PlanetOS - Stanford Engineering - Mar... por Burton Lee
Rainer Sternfeld - Planetary Big Data - PlanetOS - Stanford Engineering - Mar...Rainer Sternfeld - Planetary Big Data - PlanetOS - Stanford Engineering - Mar...
Rainer Sternfeld - Planetary Big Data - PlanetOS - Stanford Engineering - Mar...
Burton Lee2.7K visualizações
Integration for Planet Satellite Imagery por Safe Software
Integration for Planet Satellite ImageryIntegration for Planet Satellite Imagery
Integration for Planet Satellite Imagery
Safe Software5.2K visualizações
Platform for Big Data Analytics and Visual Analytics: CSIRO use cases. Februa... por Tomasz Bednarz
Platform for Big Data Analytics and Visual Analytics: CSIRO use cases. Februa...Platform for Big Data Analytics and Visual Analytics: CSIRO use cases. Februa...
Platform for Big Data Analytics and Visual Analytics: CSIRO use cases. Februa...
Tomasz Bednarz1.7K visualizações
Earth on AWS - Next-Generation Open Data Platforms por Amazon Web Services
Earth on AWS - Next-Generation Open Data PlatformsEarth on AWS - Next-Generation Open Data Platforms
Earth on AWS - Next-Generation Open Data Platforms
Amazon Web Services1.9K visualizações
Immersive 3 d visualization of remote sensing data por sipij
Immersive 3 d visualization of remote sensing dataImmersive 3 d visualization of remote sensing data
Immersive 3 d visualization of remote sensing data
sipij350 visualizações
Processing Geospatial at Scale at LocationTech por Rob Emanuele
Processing Geospatial at Scale at LocationTechProcessing Geospatial at Scale at LocationTech
Processing Geospatial at Scale at LocationTech
Rob Emanuele1.6K visualizações
From pixels to point clouds - Using drones,game engines and virtual reality t... por ARDC
From pixels to point clouds - Using drones,game engines and virtual reality t...From pixels to point clouds - Using drones,game engines and virtual reality t...
From pixels to point clouds - Using drones,game engines and virtual reality t...
ARDC1.2K visualizações
Scaling Spatial Analytics with Google Cloud & CARTO por CARTO
Scaling Spatial Analytics with Google Cloud & CARTOScaling Spatial Analytics with Google Cloud & CARTO
Scaling Spatial Analytics with Google Cloud & CARTO
CARTO190 visualizações
RemoteSensing_DeepLearning_v2.pptx por Elise Colin
RemoteSensing_DeepLearning_v2.pptxRemoteSensing_DeepLearning_v2.pptx
RemoteSensing_DeepLearning_v2.pptx
Elise Colin43 visualizações
Cyberinfrastructure and Applications Overview: Howard University June22 por marpierc
Cyberinfrastructure and Applications Overview: Howard University June22Cyberinfrastructure and Applications Overview: Howard University June22
Cyberinfrastructure and Applications Overview: Howard University June22
marpierc1.7K visualizações
GITA PNW 2015 Peter Batty por Peter Batty
GITA PNW 2015 Peter BattyGITA PNW 2015 Peter Batty
GITA PNW 2015 Peter Batty
Peter Batty980 visualizações
Soderstrom por NASAPMC
SoderstromSoderstrom
Soderstrom
NASAPMC13.9K visualizações

Mais de Codiax

Dr. Laura Kerber (NASA’s Jet Propulsion Laboratory) – Exploring Caves on the ... por
Dr. Laura Kerber (NASA’s Jet Propulsion Laboratory) – Exploring Caves on the ...Dr. Laura Kerber (NASA’s Jet Propulsion Laboratory) – Exploring Caves on the ...
Dr. Laura Kerber (NASA’s Jet Propulsion Laboratory) – Exploring Caves on the ...Codiax
206 visualizações27 slides
Costas Voliotis (CodeWeTrust) – An AI-driven approach to source code evaluation por
Costas Voliotis (CodeWeTrust) – An AI-driven approach to source code evaluationCostas Voliotis (CodeWeTrust) – An AI-driven approach to source code evaluation
Costas Voliotis (CodeWeTrust) – An AI-driven approach to source code evaluationCodiax
50 visualizações16 slides
Dr. Lobna Karoui (Fortune 500) – Disruption, empathy & Trust for sustainable ... por
Dr. Lobna Karoui (Fortune 500) – Disruption, empathy & Trust for sustainable ...Dr. Lobna Karoui (Fortune 500) – Disruption, empathy & Trust for sustainable ...
Dr. Lobna Karoui (Fortune 500) – Disruption, empathy & Trust for sustainable ...Codiax
34 visualizações42 slides
Luka Postružin (Superbet) – ‘From zero to hero’ in early life customer segmen... por
Luka Postružin (Superbet) – ‘From zero to hero’ in early life customer segmen...Luka Postružin (Superbet) – ‘From zero to hero’ in early life customer segmen...
Luka Postružin (Superbet) – ‘From zero to hero’ in early life customer segmen...Codiax
71 visualizações30 slides
Gema Parreno Piqueras (Apium Hub) – Videogames and Interactive Narrative Cont... por
Gema Parreno Piqueras (Apium Hub) – Videogames and Interactive Narrative Cont...Gema Parreno Piqueras (Apium Hub) – Videogames and Interactive Narrative Cont...
Gema Parreno Piqueras (Apium Hub) – Videogames and Interactive Narrative Cont...Codiax
26 visualizações14 slides
Janos Puskas (Accenture) – Azure IoT Reference Architecture for enterprise Io... por
Janos Puskas (Accenture) – Azure IoT Reference Architecture for enterprise Io...Janos Puskas (Accenture) – Azure IoT Reference Architecture for enterprise Io...
Janos Puskas (Accenture) – Azure IoT Reference Architecture for enterprise Io...Codiax
66 visualizações11 slides

Mais de Codiax(20)

Dr. Laura Kerber (NASA’s Jet Propulsion Laboratory) – Exploring Caves on the ... por Codiax
Dr. Laura Kerber (NASA’s Jet Propulsion Laboratory) – Exploring Caves on the ...Dr. Laura Kerber (NASA’s Jet Propulsion Laboratory) – Exploring Caves on the ...
Dr. Laura Kerber (NASA’s Jet Propulsion Laboratory) – Exploring Caves on the ...
Codiax206 visualizações
Costas Voliotis (CodeWeTrust) – An AI-driven approach to source code evaluation por Codiax
Costas Voliotis (CodeWeTrust) – An AI-driven approach to source code evaluationCostas Voliotis (CodeWeTrust) – An AI-driven approach to source code evaluation
Costas Voliotis (CodeWeTrust) – An AI-driven approach to source code evaluation
Codiax50 visualizações
Dr. Lobna Karoui (Fortune 500) – Disruption, empathy & Trust for sustainable ... por Codiax
Dr. Lobna Karoui (Fortune 500) – Disruption, empathy & Trust for sustainable ...Dr. Lobna Karoui (Fortune 500) – Disruption, empathy & Trust for sustainable ...
Dr. Lobna Karoui (Fortune 500) – Disruption, empathy & Trust for sustainable ...
Codiax34 visualizações
Luka Postružin (Superbet) – ‘From zero to hero’ in early life customer segmen... por Codiax
Luka Postružin (Superbet) – ‘From zero to hero’ in early life customer segmen...Luka Postružin (Superbet) – ‘From zero to hero’ in early life customer segmen...
Luka Postružin (Superbet) – ‘From zero to hero’ in early life customer segmen...
Codiax71 visualizações
Gema Parreno Piqueras (Apium Hub) – Videogames and Interactive Narrative Cont... por Codiax
Gema Parreno Piqueras (Apium Hub) – Videogames and Interactive Narrative Cont...Gema Parreno Piqueras (Apium Hub) – Videogames and Interactive Narrative Cont...
Gema Parreno Piqueras (Apium Hub) – Videogames and Interactive Narrative Cont...
Codiax26 visualizações
Janos Puskas (Accenture) – Azure IoT Reference Architecture for enterprise Io... por Codiax
Janos Puskas (Accenture) – Azure IoT Reference Architecture for enterprise Io...Janos Puskas (Accenture) – Azure IoT Reference Architecture for enterprise Io...
Janos Puskas (Accenture) – Azure IoT Reference Architecture for enterprise Io...
Codiax66 visualizações
Adria Recasens, DeepMind – Multi-modal self-supervised learning from videos por Codiax
Adria Recasens, DeepMind – Multi-modal self-supervised learning from videosAdria Recasens, DeepMind – Multi-modal self-supervised learning from videos
Adria Recasens, DeepMind – Multi-modal self-supervised learning from videos
Codiax101 visualizações
Javier Fuentes Alonso (Uizard) – Using machine learning to turn you into a de... por Codiax
Javier Fuentes Alonso (Uizard) – Using machine learning to turn you into a de...Javier Fuentes Alonso (Uizard) – Using machine learning to turn you into a de...
Javier Fuentes Alonso (Uizard) – Using machine learning to turn you into a de...
Codiax57 visualizações
Emeli Dral (Evidently AI) – Analyze it: production monitoring for machine lea... por Codiax
Emeli Dral (Evidently AI) – Analyze it: production monitoring for machine lea...Emeli Dral (Evidently AI) – Analyze it: production monitoring for machine lea...
Emeli Dral (Evidently AI) – Analyze it: production monitoring for machine lea...
Codiax120 visualizações
Matthias Feys (ML6) – Bias in ML: A Technical Intro por Codiax
Matthias Feys (ML6) – Bias in ML: A Technical IntroMatthias Feys (ML6) – Bias in ML: A Technical Intro
Matthias Feys (ML6) – Bias in ML: A Technical Intro
Codiax46 visualizações
Christophe Tallec, Hello Tomorrow – Solving our next decade challenges throug... por Codiax
Christophe Tallec, Hello Tomorrow – Solving our next decade challenges throug...Christophe Tallec, Hello Tomorrow – Solving our next decade challenges throug...
Christophe Tallec, Hello Tomorrow – Solving our next decade challenges throug...
Codiax33 visualizações
Sean Holden (University of Cambridge) - Proving Theorems_ Still A Major Test ... por Codiax
Sean Holden (University of Cambridge) - Proving Theorems_ Still A Major Test ...Sean Holden (University of Cambridge) - Proving Theorems_ Still A Major Test ...
Sean Holden (University of Cambridge) - Proving Theorems_ Still A Major Test ...
Codiax561 visualizações
Olga Afanasjeva (GoodAI) - Towards general artificial intelligence for common... por Codiax
Olga Afanasjeva (GoodAI) - Towards general artificial intelligence for common...Olga Afanasjeva (GoodAI) - Towards general artificial intelligence for common...
Olga Afanasjeva (GoodAI) - Towards general artificial intelligence for common...
Codiax113 visualizações
Maciej Marek (Philip Morris International) - The Tools of The Trade por Codiax
Maciej Marek (Philip Morris International) - The Tools of The TradeMaciej Marek (Philip Morris International) - The Tools of The Trade
Maciej Marek (Philip Morris International) - The Tools of The Trade
Codiax584 visualizações
Joanna Bryson (University of Bath) - Intelligence by Design_ Systems engineer... por Codiax
Joanna Bryson (University of Bath) - Intelligence by Design_ Systems engineer...Joanna Bryson (University of Bath) - Intelligence by Design_ Systems engineer...
Joanna Bryson (University of Bath) - Intelligence by Design_ Systems engineer...
Codiax172 visualizações
Jakub Langr (University of Oxford) - Overview of Generative Adversarial Netwo... por Codiax
Jakub Langr (University of Oxford) - Overview of Generative Adversarial Netwo...Jakub Langr (University of Oxford) - Overview of Generative Adversarial Netwo...
Jakub Langr (University of Oxford) - Overview of Generative Adversarial Netwo...
Codiax161 visualizações
Jakub Bartoszek (Samsung Electronics) - Hardware Security in Connected World por Codiax
Jakub Bartoszek (Samsung Electronics) - Hardware Security in Connected WorldJakub Bartoszek (Samsung Electronics) - Hardware Security in Connected World
Jakub Bartoszek (Samsung Electronics) - Hardware Security in Connected World
Codiax69 visualizações
Jair Ribeiro - Defining a Successful Artificial Intelligence Strategy for you... por Codiax
Jair Ribeiro - Defining a Successful Artificial Intelligence Strategy for you...Jair Ribeiro - Defining a Successful Artificial Intelligence Strategy for you...
Jair Ribeiro - Defining a Successful Artificial Intelligence Strategy for you...
Codiax78 visualizações
Cindy Spelt (Zoom In Zoom Out) - How to beat the face recognition challenges? por Codiax
Cindy Spelt (Zoom In Zoom Out) - How to beat the face recognition challenges?Cindy Spelt (Zoom In Zoom Out) - How to beat the face recognition challenges?
Cindy Spelt (Zoom In Zoom Out) - How to beat the face recognition challenges?
Codiax74 visualizações
Alexey Borisenko (Cisco) - Creating IoT solution using LoRaWAN Network Server por Codiax
Alexey Borisenko (Cisco) - Creating IoT solution using LoRaWAN Network ServerAlexey Borisenko (Cisco) - Creating IoT solution using LoRaWAN Network Server
Alexey Borisenko (Cisco) - Creating IoT solution using LoRaWAN Network Server
Codiax63 visualizações

Último

Info Session November 2023.pdf por
Info Session November 2023.pdfInfo Session November 2023.pdf
Info Session November 2023.pdfAleksandraKoprivica4
10 visualizações15 slides
GDG Cloud Southlake 28 Brad Taylor and Shawn Augenstein Old Problems in the N... por
GDG Cloud Southlake 28 Brad Taylor and Shawn Augenstein Old Problems in the N...GDG Cloud Southlake 28 Brad Taylor and Shawn Augenstein Old Problems in the N...
GDG Cloud Southlake 28 Brad Taylor and Shawn Augenstein Old Problems in the N...James Anderson
33 visualizações32 slides
The Research Portal of Catalonia: Growing more (information) & more (services) por
The Research Portal of Catalonia: Growing more (information) & more (services)The Research Portal of Catalonia: Growing more (information) & more (services)
The Research Portal of Catalonia: Growing more (information) & more (services)CSUC - Consorci de Serveis Universitaris de Catalunya
73 visualizações25 slides
METHOD AND SYSTEM FOR PREDICTING OPTIMAL LOAD FOR WHICH THE YIELD IS MAXIMUM ... por
METHOD AND SYSTEM FOR PREDICTING OPTIMAL LOAD FOR WHICH THE YIELD IS MAXIMUM ...METHOD AND SYSTEM FOR PREDICTING OPTIMAL LOAD FOR WHICH THE YIELD IS MAXIMUM ...
METHOD AND SYSTEM FOR PREDICTING OPTIMAL LOAD FOR WHICH THE YIELD IS MAXIMUM ...Prity Khastgir IPR Strategic India Patent Attorney Amplify Innovation
25 visualizações9 slides
Black and White Modern Science Presentation.pptx por
Black and White Modern Science Presentation.pptxBlack and White Modern Science Presentation.pptx
Black and White Modern Science Presentation.pptxmaryamkhalid2916
14 visualizações21 slides
handbook for web 3 adoption.pdf por
handbook for web 3 adoption.pdfhandbook for web 3 adoption.pdf
handbook for web 3 adoption.pdfLiveplex
19 visualizações16 slides

Último(20)

Info Session November 2023.pdf por AleksandraKoprivica4
Info Session November 2023.pdfInfo Session November 2023.pdf
Info Session November 2023.pdf
AleksandraKoprivica410 visualizações
GDG Cloud Southlake 28 Brad Taylor and Shawn Augenstein Old Problems in the N... por James Anderson
GDG Cloud Southlake 28 Brad Taylor and Shawn Augenstein Old Problems in the N...GDG Cloud Southlake 28 Brad Taylor and Shawn Augenstein Old Problems in the N...
GDG Cloud Southlake 28 Brad Taylor and Shawn Augenstein Old Problems in the N...
James Anderson33 visualizações
Black and White Modern Science Presentation.pptx por maryamkhalid2916
Black and White Modern Science Presentation.pptxBlack and White Modern Science Presentation.pptx
Black and White Modern Science Presentation.pptx
maryamkhalid291614 visualizações
handbook for web 3 adoption.pdf por Liveplex
handbook for web 3 adoption.pdfhandbook for web 3 adoption.pdf
handbook for web 3 adoption.pdf
Liveplex19 visualizações
Igniting Next Level Productivity with AI-Infused Data Integration Workflows por Safe Software
Igniting Next Level Productivity with AI-Infused Data Integration Workflows Igniting Next Level Productivity with AI-Infused Data Integration Workflows
Igniting Next Level Productivity with AI-Infused Data Integration Workflows
Safe Software225 visualizações
Transcript: The Details of Description Techniques tips and tangents on altern... por BookNet Canada
Transcript: The Details of Description Techniques tips and tangents on altern...Transcript: The Details of Description Techniques tips and tangents on altern...
Transcript: The Details of Description Techniques tips and tangents on altern...
BookNet Canada130 visualizações
【USB韌體設計課程】精選講義節錄-USB的列舉過程_艾鍗學院 por IttrainingIttraining
【USB韌體設計課程】精選講義節錄-USB的列舉過程_艾鍗學院【USB韌體設計課程】精選講義節錄-USB的列舉過程_艾鍗學院
【USB韌體設計課程】精選講義節錄-USB的列舉過程_艾鍗學院
IttrainingIttraining34 visualizações
Lilypad @ Labweek, Istanbul, 2023.pdf por Ally339821
Lilypad @ Labweek, Istanbul, 2023.pdfLilypad @ Labweek, Istanbul, 2023.pdf
Lilypad @ Labweek, Istanbul, 2023.pdf
Ally3398219 visualizações
Business Analyst Series 2023 - Week 3 Session 5 por DianaGray10
Business Analyst Series 2023 -  Week 3 Session 5Business Analyst Series 2023 -  Week 3 Session 5
Business Analyst Series 2023 - Week 3 Session 5
DianaGray10209 visualizações
Data-centric AI and the convergence of data and model engineering: opportunit... por Paolo Missier
Data-centric AI and the convergence of data and model engineering:opportunit...Data-centric AI and the convergence of data and model engineering:opportunit...
Data-centric AI and the convergence of data and model engineering: opportunit...
Paolo Missier34 visualizações
Voice Logger - Telephony Integration Solution at Aegis por Nirmal Sharma
Voice Logger - Telephony Integration Solution at AegisVoice Logger - Telephony Integration Solution at Aegis
Voice Logger - Telephony Integration Solution at Aegis
Nirmal Sharma17 visualizações
Web Dev - 1 PPT.pdf por gdsczhcet
Web Dev - 1 PPT.pdfWeb Dev - 1 PPT.pdf
Web Dev - 1 PPT.pdf
gdsczhcet55 visualizações
Kyo - Functional Scala 2023.pdf por Flavio W. Brasil
Kyo - Functional Scala 2023.pdfKyo - Functional Scala 2023.pdf
Kyo - Functional Scala 2023.pdf
Flavio W. Brasil165 visualizações
From chaos to control: Managing migrations and Microsoft 365 with ShareGate! por sammart93
From chaos to control: Managing migrations and Microsoft 365 with ShareGate!From chaos to control: Managing migrations and Microsoft 365 with ShareGate!
From chaos to control: Managing migrations and Microsoft 365 with ShareGate!
sammart939 visualizações
Report 2030 Digital Decade por Massimo Talia
Report 2030 Digital DecadeReport 2030 Digital Decade
Report 2030 Digital Decade
Massimo Talia14 visualizações
Special_edition_innovator_2023.pdf por WillDavies22
Special_edition_innovator_2023.pdfSpecial_edition_innovator_2023.pdf
Special_edition_innovator_2023.pdf
WillDavies2216 visualizações

Roelof Pieters (Overstory) – Tackling Forest Fires and Deforestation with Satellite Data and AI

  • 1. Roelof Pieters CTO & Co-founder roelof@overstory.ai @graphific Tackling Forest Fires and Deforestation with Satellite Data and AI May 14, 2021
  • 2. • Computer Science/Deep Learning • Anthropology and Development Sociology • Previously founded Creative.ai, Stockholm.ai, GitXiv and other open source initiatives/networks/startups Introduction Roelof Pieters CTO & Co-founder, Overstory This session • Advancements in satellite data & AI • Example: deforestation monitoring • Example: preventing wildfires and power outages
  • 3. Indra den Bakker CEO & Co-founder Anniek Schouten COO & Co-founder David de Meij Data Scientist Roelof Pieters CTO & Co-founder Rochelle Silva Radar & GIS Researcher Lorenzo Riches Data Scientist We’re Overstory Arsha Yuditha Amiranti GIS Specialist Elvira Garkava Business Developer Fiona Spruill Chief Product Officer Andrea Giardini DevOps Engineer Killian Tobin Head of Business Development
  • 5. The problem Insect outbreaks Forest fires Deforestation Forest fire (Source: Nasa) Bark beetle infestation (Source: Sodra) Deforestation (Source: ESA, Google)
  • 7. Satellite market Today 2029 # of satellites 2010 1000 800 600 400 200 1,470 satellites launched between 2010 and 2019 8,500 satellites expected to launch between 2019 and 2028* Hyperspectral VHR SAR Video Real-time Source:Euroconsult: https://spacenews.com/analysis-are-smallsats-entering-the-maturity-stage/
  • 8. Input data High-resolution satellite imagery up to 30 cm resolution
  • 9. Input data Invisible light like near-IR Multispectral data
  • 10. Input data Weather independent radar images SAR data
  • 11. July 2, 2017 July 4, 2017 July 5, 2017 Input data Frequent revisits up to daily on a global scale
  • 12. Input data Frequent revisits up to daily on a global scale 2.5m source: https://platform.digitalglobe.com/earth-imaging-basics-spatial-resolution/ 30cm
  • 13. Tri-stereo and video monitoring for 3D-mapping SAR data to look through clouds Multi- and hyperspectral data Up to 30 cm resolution Up to 20 VHR daily revisits & geostationary satellites Very high-resolution satellite data
  • 14. ● Mega Large Imagery ● Constant change ● Noisy satellite data Data Science Challenges ● Noisy or lack of labels ● Generalize from training ● “ground truth” ● Classical machine learning still the norm
  • 15. Unsupervised Machine Learning ??? A? B? C? Allows Overstory to get insights anywhere in the world with 1-2 factors less customer/labeled data (1000 instead of 10K-100K data points) "Everything is related to everything else, but near things are more related than distant things"
  • 16. Stereo Imaging For creating height maps and 3D maps we use Stereo Imaging techniques to create Digital Surface Maps (DSM) and Digital Terrain Maps (DTM), as well as 3D point clouds as in this example
  • 17. Stereo Imaging Learning Feature Matching with Graph Neural Networks Magic Leap/ETH Zurich (2020)
  • 18. A process by which freely accessible low resolution imagery can be upscaled to commercial-grade high resolution, allowing for more accurate insights, easier labelling by our annotators, and increased accuracy for our machine learning models, at a cheaper cost (Upscaling Landsat-8 to Digital Globe Worldview-3-like level) LS8 TIRS LS8 CIRRUS LS8 Panchromatic LS8 SWIR LS8 NIR LS8 RGB DG-WV3 RGB DG-WV3 Panchromatic DG-WV3 SWIR 31cm 1.24m 3.7m 15m 100m 30m Generative Upscaling Li et al (2019) Feedback Network for Image Super-Resolution" See also Super-Resolution Generative Adversarial Network(s) (many papers)
  • 19. Active Learning (re)train candidate selection oracle / human annotator ● eg BAyesian Active Learning library (BaaL) by ElementAI: https://github.com/ElementAI/baal/ ○ MCDropout (Gal et al. 2015) ○ BALD (Houlsby et al. 2011) (ElementAI)
  • 20. Living data repository training data for scalable and accurate deep learning algorithms
  • 21. case #1: Machine Learning for deforestation monitoring
  • 22. ● high resolution forest and landcover map ● up to date with 2014 to now ● high forest and crop type accuracy Deforestation monitoring what how ● deep learning for segmentation ● multiple satellite sources: sensor fusion / multimodality ● noisy data: generative gap filling ● dynamic data regime: open data, active labelling (ground and satellite), noisy labels
  • 23. Climate & weather Height data Advancements in deep learning SAR (radar) data Multi resolution data Sensor fusion Convolutional Neural Networks Unsupervised learning and active (bayesian) learning Multispectral data Output Segmentation Layers of satellite imagery Land Cover Segmentation on pixel level
  • 24. Active Learning Encoder Neural Planet Embedding 1. 2. 3. 4a. Data repositor y 5. 4b. (Overstory training data pipeline) Annotator Field worker Public Data
  • 27. High-Resolution Global Maps of 21st- Century Forest Cover Change (Hansen et al., Science 2013).
  • 28. Overstory 10m and less Hansen et al. 2013/2019 1.7 (2000) GlobCover 2010 2.3 (2009) CCI Landcover Africa 20m (2016)
  • 29. case # 2: Scaling ML for very high resolution earth observation & risk monitoring
  • 30. Vegetation & powerlines of global annual CO2 emissions ~1% direct economic losses $21B people without power 40M
  • 33. ● very high resolution tree species map: 37 different tree species, shrub species and grass classes ● over 16,000 km of power lines with a corridor width of 150 meter ● total area of over 505,000 km2 ● hard constraints on minimum level of accuracy fire risk monitoring what how ● deep learning, naturally :) ● open source labels, customer labels, external party ground validation ● very high resolution satellite imagery (50cm) ● massively parallel distributed processing through dask, kubernetes, and distributed data parallel training
  • 34. Jupyterhub Jupyterhub allows users to create dedicated computational environments Our Infrastructure
  • 35. Getting a new notebook Our Infrastructure - Jupyterhub 👥 I need a new notebook! Processing... We need a new machine for this... New node Notebook available 1 2 3 4 5 6
  • 36. Dask Dask provides advanced parallelism for analytics, enabling performance at scale for the tools you love Our Infrastructure
  • 37. I need some heavy compute Our Infrastructure - Dask 👥 I need a new Dask cluster! Dask-gateway Dask-scheduler Dask-worker Dask-worker Cluster available 1 2 3 4 5 6
  • 38. Dask cluster - Structure Our Infrastructure - Dask Dask-gateway Dask-scheduler Dask-scheduler Dask-worker Dask-worker Dask-worker Dask-worker Dask-worker Dask-worker Dask-worker Dask-worker Dask-worker Dask-worker Dask-worker Dask-worker
  • 39. Papermill Papermill provides queues and ETL capabilities for running and logging jupyter notebook workloads Our Infrastructure
  • 40. Our Infrastructure - Papermill Image by Matthew Seal, Netflix
  • 41. Our Infrastructure - Papermill Image by Matthew Seal, Netflix
  • 42. All together now Our Infrastructure Abernathey et al. (2021): Pangeo architecture diagram.
  • 46. Real-time Vegetation Intelligence Platform Tree height Canopy segmentation Fire risks Tree species Verification of trimming Storm damage Risk identification Predictive prioritization
  • 47. Overstory is on a mission to monitor all natural resources on Earth in real-time.
  • 48. We’re in this together! Overstory is on a mission to monitor all natural resources on Earth in real-time. Come join us! https://www.overstory.com/careers
  • 49. Learning Resources / Research ● List of videos about Geospatial data science from FOSS4G (Free and Open Source Software for Geospatial / OSGeo) https://www.youtube.com/channel/UC_2Lyc9VUX-jC-E1prJitHw/vi deos ● ICLR 2020 proceedings now available: https://iclr.cc/virtual_2020/ & video recordings of climate change AI workshop: https://www.youtube.com/channel/UCyjDr_aoMlzhSvCTdT7eZ9g/ videos ● CVPR pre-papers for EarthVision: Large Scale Computer Vision for Remote Sensing Imagery Workshop http://openaccess.thecvf.com/CVPR2020_workshops/CVPR2020 _w11.py Awesome lists of resources: ● https://github.com/sshuair/awesome-gis ● https://github.com/robmarkcole/satellite-image-deep-learning ● https://github.com/chrieke/awesome-satellite-imagery-datasets ● https://github.com/acgeospatial/awesome-earthobservation-code ● https://github.com/wenhwu/awesome-remote-sensing-change- detection Amazing projects ● https://www.globalforestwatch.org/ ● https://trase.earth/ ● https://www.half-earthproject.org/ ● https://www.climatewatchdata.org/ Geospatial Toolkit/UI: ● QGIS: https://www.qgis.org Satellite Data ● sentinel-2 10m resolution satellite imagery: https://scihub.copernicus.eu/dhus/#/home ● landsat 30m resolution satellite imagery: https://landsat.gsfc.nasa.gov/ ML for geospatial/satellite data libraries: ● https://github.com/sentinel-hub/eo-learn ● https://rastervision.io/ ● https://github.com/fastai/fastai2/ Resources
  • 50. Rolnick, et al. Tackling Climate Change with Machine Learning, arXiv:1906.05433 & https://www.climatechange.ai/