Solve Geosolutions and DiUS Partner to Transform the Mining Industry Using Artificial Intelligence on AWS

DiUS is an APN Advanced Consulting Partner with an AWS Machine Learning Competency

Helping Mining Companies Limit Manual Tasks and Become Data-Driven

Improvements in imaging technology enable mining companies to capture, store, and analyze drill-core imagery in new, innovative ways. Technological advances also help them improve operational efficiencies by using better imagery, collecting many more data points, and revisiting historical data. However, these new capabilities can also require more time and manual resources, particularly from geologists, who must sift through a more significant and complex scope of data. Many mining companies therefore must decide if reinventing themselves in a data-driven world is a worthwhile effort. For the team at Solve Geosolutions, the decision is yes.

Through education, collaboration, and advanced tooling, the company’s mission is to help mining companies and geologists take advantage of new data analytics capabilities to drive insights, operational improvements, and, ultimately, pursue new business outcomes. Solve Geosolutions seeks to change the way mining companies use technology, such as machine learning, to limit geologists’ manual inspection tasks of drill core and help them focus on the heart of what they do: using their domain knowledge to understand and interpret geology.

“We have a deep background in machine learning across our team. Our mission is to be at the cutting edge of machine learning to solve geological problems,” says Brenton Crawford, director and geoscientist at Solve Geosolutions. “We want to change the way geologists work. Our goal is to help them make the most of the geological data.”

“We’re thrilled to be working with DiUS to help mining operations take advantage of creating new value through artificial intelligence on AWS.”

- Liam Webb, Director at Solve Geosolutions and Chief Executive Officer at DataRock

Using Machine Learning and Working with DiUS to Change the Way Mining Companies Approach Data

Already an expert in the application of machine learning to solve geological problems, Solve wanted to build an automated and scalable solution to help mining customers tap into increasingly complex sources of data and objectively analyze large repositories of drill core imagery. The company sought to explore the application of image processing algorithms to high-resolution images and automate the extraction of geological information to deliver more consistent results.

The company sought to use deep learning to help companies generate new insights at scale. However, the team felt they were missing some core skills, like deep learning expertise in cutting-edge technologies, to develop and launch such a model. The team decided to build its solution on Amazon Web Services (AWS) and reached out to AWS for consulting partner recommendations. AWS quickly brought DiUS into the conversation.

DiUS is an APN Advanced Consulting Partner and Machine Learning Competency Partner that was founded by two individuals passionate about innovation and taking advantage of emerging technologies. “Almost everything DiUS does is cloud-native and on AWS,” says Frank Losinno, client engagement principal at DiUS and chief technology officer at DataRock. “We always have the AWS Cloud underpinning how we develop products and solutions.” With a strong focus on machine learning, specifically deep learning analysis, and proven expertise in AWS technology, DiUS provided a strong complement to Solve’s expertise in analyzing and understanding exploration and mining data. 

The Solve team began working with DiUS to build a smaller-scale proof of concept for its solution. “We turned to DiUS to help us build a model using deep learning to help with complex image processing to extract geological information from core imagery. Anything we asked DiUS to do, they did. It felt anything was possible from the beginning,” says Liam Webb, director at Solve Geosolutions and chief executive officer at DataRock. “Working with a highly professional, well-run, agile, and flexible company like DiUS makes a big difference in our ability to accomplish our desired outcomes.”

Creating a Solution to Streamline Image Processing and Drive Faster Results for Geologists and Mining Companies

Together, DiUS and Solve developed a solution to apply a range of deep learning-based image analysis techniques to extract geological information from drill core imagery.

Within weeks, DiUS and Solve built a deep learning model on AWS to extract a specific geological element requested by a client. The solution used Amazon Simple Storage Service (Amazon S3), a deep neural network architecture called Mask R-CNN, and Amazon Elastic Compute Cloud (Amazon EC2) P3 Instances to accelerate machine learning and high-performance computing applications. DiUS also used AWS Deep Learning AMIs with Open Source Frameworks and implemented its model in TensorFlow, hosted as a docker container on AWS Fargate. The application to prepare the dataset and apply batch predictions was a static react frontend served from an S3 bucket using Amazon CloudFront and Amazon Cognito for authentication.

The results of the segmentation models were really impressive. “We very quickly found we were able to detect important geological features in images at a scale, resolution, and consistency impossible to collect manually by a geologist," says Webb.

Taking Innovation to the Next Level with DataRock, a SaaS Solution on AWS

Following the successful deployment of its models and several new project leads, both companies came together to discuss the possibility of developing a joint venture and business plan to launch a new end-to-end solution to market unlike anything currently available. Through their joint venture, the companies could focus on building a commercialized product that could help mining companies take advantage of image segmentation technology to automatically analyze drill core imagery and deploy machine learning models to help drive more significant insights and better decision-making in the field.

The result of the venture is DataRock, a machine learning-powered, cloud-based image analysis platform specifically designed for exploration and mining. The startup’s goal is to help mining companies be more flexible and agile in where they drill while saving time and money to put towards other business-critical projects.

“In the face of increasing industry challenges such as low commodity prices and declining grades, DataRock provides accurate, fast, and consistent high-resolution information about a mineral deposit’s geology that aids decision making throughout the entire mining cycle, delivering important productivity and throughput savings to a mine’s bottom line,” says Webb. For example, it may have previously taken a company up to six weeks to obtain geo-chemical assay results from a mining site. Using DataRock, companies can get proxies within hours. This near real-time power enables more intelligent decision-making regarding the viability of a particular mining site.

DataRock is using many AWS services to power its platform, including Amazon SageMaker to build and deploy its ML models, Amazon Simple Queue Service (SQS) to orchestrate the process, AWS Lambda for processing and serving the UI, Amazon API Gateway for the REST API, and Amazon S3 for storage of core image sets. “We are also currently working with the AWS GroundTruth team to collaborate in building out feature sets for our workflow and product needs,” says Losinno. 

Having recently launched DataRock, the Solve and DiUS teams are just beginning to go to market and educate mining companies around the globe about the benefits of the DataRock solution. “We’re thrilled to be working with DiUS to help mining operations take advantage of creating new value through artificial intelligence on AWS,” says Webb.

Solve Geosolutions Logo

About Solve Geosolutions

Solve Geosolutions is a data science consultancy based in Melbourne. Founded in 2015, the company specializes in statistical analysis and machine learning-based solutions for exploration and mining. Solve’s aim is to develop and implement data science-based solutions that improve the way companies mine and explore large and complex multivariate datasets. In addition to designing end-to-end solutions, Solve strives to empower its clients to think about their data in new and innovative ways through collaborative engagement and training.

Challenge

Already an expert in the application of machine learning to solve geological problems, Solve Geosolutions wanted to build an automated and scalable solution to help mining customers tap into increasingly complex sources of data and become more data-driven in their decision making. However, Solve did not have the expertise it needed to build the type of deep learning models it sought to develop using AWS.

Solution

The company decided to work with DiUS to build a solution on AWS. Together, DiUS and Solve worked to develop a solution to apply a range of deep learning-based image analysis techniques to extract geological information from drill core imagery. Within weeks, DiUS and Solve had combined forces to build a deep learning model on AWS designed to extract a specific geological element requested by a client.

Benefit

Both companies quickly found that using AWS, they were able to detect important geological features in images at a scale, resolution, and consistency impossible to collect manually by a geologist. The teams decided to explore a new line of business and form a joint venture to launch a software as a service (SaaS) product to market called DataRock, a machine learning-powered, cloud-based image analysis platform specifically designed for exploration and mining.

About DiUS

DiUS is an Australian technology services company that helps organizations both large and small to solve hard problems, get new ideas to market, and disrupt and transform business models. The company strives to deliver sustained success by delivering a platform for growth. Its services span the entire product lifecycle, from ideation and service design and PoCs to creating an MVP and building a product to scale. DiUS’ end-to-to end software and hardware capabilities, coupled with its ability to explore emerging technologies for commercial application, make it unique in the market. DiUS is an APN Advanced Consulting Partner and holds the AWS Machine Learning Competency.

Published December 2019