Machine Learning + Synthetic Data

The smart, fast and affordable way to detect defects in critical infrastructure.

The problem we solve

Synthetic Data Pty Ltd develops and distributes software that helps asset managers and engineers use the accelerated speed and deep learning properties of machine learning to detect defects in critical assets quickly, safely and cost effectively.

Can machine learning be used in asset management to prevent critical assets from failing?
We think so.

The benefits

 Increase the speed of asset inspections

 Increase the frequency of asset inspections

 Increase the compliance with industry regulation

 Decrease the cost of asset inspections

How we do it

We use machine learning to replace the manual inspection of infrastructure assets with automated inspection. This enables defects to be readily identified and repaired before they cause failures. This form of predictive asset maintenance can save your organisation potentially millions of dollars, and protect the lives of those who maintain your assets.

Who we work with

Our software can create synthetic photos of almost any asset and any defect and create the vast volumes of training data you need to train the model. From power poles and pipelines to mining and medical imaging, we’ve got you covered. Put it this way: if you can see it, we can synthesise it.

How we help utilities automate their
visual inspection process:

Take a look at our pitch video and discover what Synthetic Data (formerly known as Machine Dreams) can offer, who we work with, how our software can save utilities millions in inspection costs and how we’re helping asset engineers automate their visual inspection process.

View the slide deck of the pitch video:

A cautionary tale:

How machine learning can be used to prevent catastrophic bush fires. 

The 2019 bush fires in California were reported to have been caused by defective power assets. These fires killed hundreds of people, left the power firm with billion-dollar litigation bills and created a significant loss of trust in their social licence to operate. 

Our secret sauce

We’ve developed an exciting piece of simulation software that generates the millions and millions of photos/data needed to train the computer to spot the defects. This data is known as ‘synthetic data’ and it will revolutionise the way machine learning is conducted in the future. We provide this missing piece of the machine learning puzzle that makes machine learning accessible and affordable for all.

About Us

A finalist in the 2020 Amazon Web Services (AWS) Machine Learning Pitch Event
A finalist in the 2020 Australian Technologies Competition Award.
The winner of the 2019 National Asset Management Conference Startup Pitch Competition.

Our Team

Bernadette Schwerdt



Bernadette is a marketing strategist with 30 years’ experience in the sector. She was a senior director with one of the world’s largest advertising agency, Young & Rubicam and created campaigns for global giants including Apple, American Express and BHP. She has a degree in Business (marketing) from the University of South Australia, and has been a lecturer in marketing, entrepreneurship and digital marketing at the University of Melbourne. Melbourne Business School, and RMIT. She is the author of three best-selling business books (Secrets of Online Entrepreneurship, How to Build an Online Business and Catch of the Decade).  She was listed as one of the Top 50 Small Business Leaders Award in 2020 by Inside Small Business Magazine and is a popular TEDx speaker.

Greg Baker



Greg is an experienced technologist and entrepreneur with two successful exits. He has a Bachelor of Science and Mathematics from Macquarie University and is close to completing his PhD in  Cryptography from Macquarie University.  He began his career by writing the world’s first quantum computing simulator. He has advised start-ups and Fortune 500 companies, and worked with governments in five countries. He wrote the terms for digital rights management in the USA-Australian free trade agreement and was asked to give parliamentary evidence on Wi-Fi broadband policy. Past employers include CSIRO, Google and Atlassian and was a director at the Institute of Open Systems Technology that worked extensively to provide Australian startups with funding, mentorship and export opportunities.  

Why is machine learning so difficult to access?

Most machine learning experts know you need millions of photos to train the model to detect a specific detect. But who has the millions of photos needed to train the model? We do. 

We use game engine technology to create ‘synthetic’ photos of your asset and defective asset so we can generate the vast volume of training data needed to train the model. 

Pretty cool huh? 

And best of all, our photos are automatically labelled with training data, eliminating the costly and time-consuming step of labelling data. That one step alone can save you literally millions. 

 Think you can’t afford machine learning because you don’t have the data you need to train the model? Think again. Our software creates the data when there is none and makes machine learning accessible and affordable for us. We’ve seen the devastation that occurs when big assets fail in a big way. We’d like to work with you to ensure that yours don’t.

Our Process:

Whilst machine learning can be a complex process, our process of applying it in your business is simple. Take a look at our 6-step process of how we take your photos of assets and defects and turn them into the millions of synthetic photos needed to train the model. Our secret sauce is our simulator. It generates the vast volume of photos needed to train the model.

Here’s our 6-step process to fast tracking the automated inspection of your assets:

How much can this process save?

Use case:

Power Poles: Here's a summary of the vast savings we can offer a power company

Using synthetic data to create machine learning models, we can help companies:

 increase the frequency of asset assessment

 increase the rate of defect detection

 reduce the cost of compliance

 reduce the risk of defects causing disasters

 reduce insurance premiums

 reduce risk of management being exposed to criminal liability/workplace manslaughter

The assets we inspect

The defects we detect

The defects we detect

- Worn ball bearings
- Rusty conveyor belts

- Peeling paint
- Defective cladding

- Rusted pipes
- Cracked concrete

- Cracked insulators
- Broken ties

- Ore sample analysis
- Objects in slurries

- X-ray scans
- Screening

Contact Details

Ph:  0419 891 932

530 Little Collins St Melbourne 3008