2. Definitions
• CAD – Computer aided design, a commonly used engineering approach creating
digital models of the physical assets.
• Physical assets – objects such as equipment, vessels, vehicles, buildings, other
installations, infrastructure etc.
• CAD models – digital engineering models of the assets in CAD formats
• Legacy assets – assets created prior to 4th industrial revolution. Their engineering
data is often contained only in paper (or scanned) drawings
• Digitising of legacy assets – building CAD models of those assets
• SaaS – software as a service
• Automated documents understanding – an AI-powered solution to recognise the
meaning of the scanned document, use this information for the business purposes
and convert it into other required formats. This process is also a part of Robotic
Process Automation (RPA) technology.
3. 1 page summary
1. Dii develops SaaS product line for automated understanding of
engineering information.
2. We enable automated creation of CAD models of the physical assets.
3. Our method addresses global industrial challenge – digitising legacy assets
4. We believe we can unlock a multi-billion dollar potential of digital
technologies affected by this challenge and become a standard platform
for resolving this challenge.
5. Our AI-powered method is innovative and original. It is proven in closed
testing and is on the way to industrial application.
6. Our team is 2 co-founders and a group of engineers with years of relevant
experience.
7. Our first product is ‘Reader’ focused on automated understanding of
scanned engineering drawings.
8. We have started to offer our service to the market bringing much higher
automation, speed and reduced cost.
5. Digital management of the physical assets
An underlying strategic trend*
• Physical assets are now managed by various technological applications all
though their lifecycles.
• Known applications include: digital twins, smart cities; smart homes;
automated industrial facilities, product lifecycle management solutions etc.
• Digital twin technology was named for several years by leading analysts as a
key global strategic trend.
• These are newly established multi-billion industries with huge growth
potential.
• All of them require CAD models of the objects.
* See more in a special deck: https://dii.ai/wp-content/uploads/2020/12/DII-P1.pdf
7. Re-using the old assets is a key idea for
circular economy
• Circular economy is one of the most
prospective emerging economic trends
in the world
• Reusing complex objects requires their
re-engineering, which itself requires
conversion drawings from paper to CAD
models.
• Thus, digitising legacy assets is a
requirement for sustainable
development
Sustainability dimension
8. The challenge – digitising legacy assets
• Most of the global assets were created before digital transformation.
• Even if they were designed in CAD, the files are often not preserved.
Most of engineering data for 15+ years old assets are kept in paper
form (scanned at best).
• Object’s 3D laser scanning or photographing provide very limited
information.
• Converting paper/scanned drawings information to CAD is largely
manual requiring substantial time and effort.
• There is no way humanity can manually digitise all important assets; an
automation solution is required.
9. Cambridge University’s Centre for Digital Built
Britain called digitisation of legacy assets
challenge a main roadblock for digital assets
management technology adoption
10. Existing solutions. Not enough.
• Scanning of the technical drawings gives a raster image; essentially a
bunch of dots. Raster-to-vector conversion only turns dots into lines
with no meaning, interconnection or layers.
• 3D laser scanning of the actual asset provides only surface information
and is very limited. Similar situation is with photographing the object.
• All other work (over 90%) is manual, requiring time/effort.
• Engineering firms offer manual work with some small degree of
automation.
• Academic research on automated drawings understanding took place
since 1980s but did not offer commercial solution yet. However it still
takes place (UK, China etc.).
• Today AI provides previously unavailable solutions for this issue.
11. Our goal: automate 90%
• We aim to build a standard set of solutions offering high rate of
automation combing various input data and methods.
• We aim to build a standard set of solutions offering high rate of
automation combing various input data and methods.
• Our solution will
unlock the true
massive potential
of above-
mentioned
technologies and
will accelerate
their adoption.
12. Team
A team of six engineers is currently
working on the project full-time assisted
by a group of project advisors.
Founders:
• Andrew Zagorodnyuk, co-founder, ideologist, investor and CEO of the project. 20+ years
of management experience running industrial and construction engineering teams.
Previous projects included highly complex engineering solutions with digital prototypes
and calculations. Customers included large international industrial groups.
• Roman Zakharov, co-founder, main tech partner, head of R&D. Mathematician. Studied
and worked in Ukraine and Western Europe. 10+ years of experience in AI and machine
learning. Co-founder of the company, who has been working for more than 7 years in
computer vision in the B2B markets and serves large international clients.
More on our team: https://dii.ai/team/
13. Part 2 of 2: The Reader
Demo: https://www.youtube.com/watch?v=3oDvoVjb2bY&t=56s
14. • Our first product is the Reader. It addresses the key part of assets digitisation –
converting the scanned drawings into CAD models.
• Reader works as a SaaS solution and recognizes objects by processing scanned paper
engineering drawings, reads their parameters and creates a CAD model using
obtained information
15. Unique algorithm
• We have developed a working algorithm able to reconstruct the raster
drawing image into a layered CAD file. It is implemented as API
available for the internal testers group, which soon will be available to
the wider community.
• The innovative part of it is that we recognize what’s is actually drawn,
understanding the meaning of vector primitives. Contrary to the simple
raster to vector conversion we split data into separate layers (symbols,
technical lines, the object itself, etc.), recognize symbols and digits to
assign the actual dimensions to the object. Such a delivery allows an
engineer to significantly reduce a scan to CAD conversion burden.
Instead of drawing it on top of vectorised raster representation,
engineer does only a certain amount of fine-tuning
16. Current state of development
Lab model (proof of the concept) - done
Closed testing – done (level of automation
reached is min 50%)
Open testing and commercial services – From
Sept 2021
• The Reader is now a fully functional prototype,
which converts drawing images into valid CAD
files with data split into separate layers. Now it
works well on mechanical and structural
drawings.
17. Pathway to a market leadership. Steps:
STEPS Degree of
automation
Service delivery
0. Current market offering. Scan-to-CAD conversion is mostly
manual.
10% Result is delivered as an
engineering service.
1. Dii Reader. Manual work is still required but for most of the
drawings* automation is much higher.
50% Result is delivered as
engineering service but
much quicker and cheaper
enhanced by software.
2. Dii Reader is trained on various types of drawings, has
consumed large dataset and is released as product available for
mass market.
70% Result is delivered as SaaS
with supportive service.
3. Dii Reader passed the learning curve and is commonly used by
various industries as a standard solution.
90% SaaS
* mechanical and structural drawings now
18. Going to market.
From August 2021 Dii started to work with prospective customers offering drawing
conversion services powered by Reader. Key advantages: much higher speed and
lower cost. Initial reaction of the market is very positive. Industries targeted:
1. Legacy assets digitisation. Converting old drawings for infrastructure and
products into CAD for automation purposes. For architecture and manufacturing.
2. Industrial decommissioning. Safe and efficient decommissioning of industrial
facilities requires engineering phase. It needs CAD files of the objects.
3. Smart buildings and smart cities. Digital solutions for built environment require
substantial conversion for legacy assets.
4. Digital twins. Huge potential market requiring assets digitisation solutions.