#StandardsGoals for 2024: What’s new for BISAC - Tech Forum 2024
Can We Trust AI for Sat Imagery Processing
1. Taras Matselyukh, CEO and founder of OPT/NET BV
Bridging the gap Between Technology & Humans
Can we trust AI for EOD processing?
In collaboration with ESA and Copernicus Data Hub
2. Seva M.
Business Dev. Director
past: BD/Account Manager
Twona/Pharmaceutical Design
Mohan R.
Marketing Director
Co-Founder: Core Digital
Strategy, Kane & Rao Group
Taras M.
CEO/Director
20 year+ industry Experience
past: Cisco/Juniper
Erik D.
CCO
past: CEO Stack State
VP Sales EMEA Digital Guardian
Gianluca V.
CFO
Co-Founder: Equidam
Founder: Metrics Specialist
Vlad Z.
Software Dev. Team Lead
Backend systems
Big-data visualisation
Business Acceleration
Partners
OPT/NET BV TEAM
3. ACCOLADES
Innovation Summit
Winner
EC Security Challenge
2017 Winner, 2018 Finalist
Hello Tomorrow
Top 500 Globally
Accenture Innovation
Awards
Top 25 - Living & Working
Business and
Technology
Partners
The AIconics
awards, 2018
San Francisco AI Summit
Best Innovation
in RPA, shortlist
4. just a clever way of doing linear algebra?Super power that can destroy us all? or
WHAT IS AI?
5. MOST POPULAR TYPES OF
AI FOR EOD PROCESSING
ML
Supervised
Unsupervis
ed
Structured
environmen
ts
Needs
labeled data
Unstructure
d
environmen
ts
Needs good
model
13. Satellite collects imagery
(SAR, SLSTR, etc.)
AI Engine analyses
SAR data in real time
AI Engine autonomously
clusters & alarms Operators
in development AI Engine
actually detected
the ‘eye of the storm’ footprint
on the ground and sea
as we monitored the
Florence Hurricane on
Sept 14 2018
while it hit East cost of US
14. OPTOSS AI GEO
SOLUTION
100K developers from all over the
world
2500+ applications developed
Semifinalist - among top 10 solutions
Concept Architecture for Flood Detection
Team OptOSS AI GEO
Architecture for Flood Detection
Team OptOSS AI GEO
Copernicus open data portals
Operator selects imagery and pre-processes the SAT data,
then pushes the Docker image into IBM Cloud for processing
docker containers are executed
in IBM Cloud on demand
AI detects flooded
areas and storm
footprints
processed maps
are stored in COS
as COGTIF
Web app serves the data
to the users of the service
Users of the service monitor the
flooded areas in the web browser
15. OptOSS AI Satellite
Observation GUI
COPERNICUS
SECURITY
CHALLENGE WINNER
Ivan Konaktchiev, Policy Officer European
Commission
The jury selected OPTOSS as the
winner of this challenge because
it deals with large amounts of
observational data under stressful
conditions.
“
”
16. CONTACT
@OptNetConsBV
+31 (0) 72 581 5940
taras@opt-net.eu
Managing the chaos since February 2018
OPT/NET
BV
http://www.opt-net.eu
17. HOW DOES IT WORK?
1.Data
Streamed into
OptOSS AI
2.Real
time
Analysis
3.Autonomou
s Clustering
4.Educatio
n by
Analyst
5.Appropriat
e
Response
6.Knowled
ge
Recycle
Editor's Notes
Good afternoon. I am Taras Matselyukh, CEO and founder of OPT/NET
ID 180886: TSAR AI - rapid and precise surface change detection
Our team is young but experienced with proven track record.
OPT/NET is working with potential users and user-clients already and expanding our service portfolio:*We have an expressed interest from EY Innovation department to jointly develop new solutions in Security market segment.
During the 2018 World Government Summit in Dubai LAST week (Feb 13), Google X co-founder Sebastian Thurn stated that artificial intelligence innovations would usher in an era of “superhuman” workers who are capable of doing far more work than what is conventionally possible.
OPTOSS AI is doing it NOW! We give engineers of telecoms “super human” powers!
Collecting and possessing data alone is not enough … the real challenge is to assess risks, predict affected areas and detect damage in very short time, as close to real-time preferably.
This task is similar to finding a proverbial needle in a haystack. Limited time to response makes this task impossibly difficult.
Can human operators solve this challenge?…
As the disaster threat continuously increases along with the sheer volume of EOD, the human capability to deal with the challenge remains the same. It takes time, money, and effort to train satellite imagery analysts, but historical performance illustrates that
costly and tragic mistakes are still being made by authorities when alerting populations about local disasters and emergencies[3].
These mistakes are root causes in more than 60% of all incidents.
Short time to make the right decisions multiplies the effects of wrong choices - which lead to losses.
Our conclusion is that current practices are insufficient for timely reaction and prevention of such situations.
Our response to the problem is an automatic AI assisted disaster monitoring and alerting system.