Mechanical Engineers Take Out First Prize at Unearthed Hunter Valley

During the last weekend of August, Team Arnut took out first prize at Unearthed Hunter Valley.

Contrary to the typical hackathon entry, team members had been exposed to limited programming throughout their mechanical engineering studies.

All three members of the team are active researchers within the Centre for Bulk Solids and Particulate Technologies (CBSPT) at the University of Newcastle’s Institute for Energy and Resources (NIER).

The team tackled the NSW Government’s Automate image analysis of Hylogger drill core data challenge, by developing an automatic image segmentation algorithm to identify the main features in RGB drill core images.

Team Arnut Victorious – First Prize Winners at Unearthed Hunter Valley: Wei Chen, Timothy Donohue and Jeremy Miller with judges Jacques Labuschagne, Derek Lush, Trent Bagnall and Chris Yeats

BACKGROUND

Team Arnut is comprised of Tim Donohue, Wei Chen and Jeremy Miller.

Tim completed his mechanical engineering degree in 2003, his PhD on the topic of permeability in porous materials in 2008, and has since been working as a consulting engineer with TUNRA Bulk Solids (TBS).

Wei completed his metallurgical engineering degree in 2009, his PhD on material handling in 2013, and has been working as a Research Associate in The University of Newcastle.

Jeremy is currently completing his mechanical engineering degree whilst working as a Research Assistant at The University of Newcastle.

EXPERIENCE

The team came into the challenge with a preconceived idea of what their solution might be, however after their Friday night presentation, they discovered they may need to change tack.

Nonetheless, their approach was to firstly understand the data and the corresponding limitations of the data.

“We spent considerable time investigating the data and images as well as considering possible approaches. The final approach we used was developed Saturday morning, and from there we had the base of the solution we developed” – Tim Donohue.

They had some previous experience with Matlab and image analysis, so these were the skills they combined with some statistical analysis of the Hylogger data.

It was crucial to have a good understanding of the challenge, plus the relevance and potential benefits to industry, which they obtained through discussions with the mentors, along with their own personal experience.

RESULTS

The award winning prototype is an automatic image segmentation algorithm, planned and developed to identify the dominant features, defined by RGB values, on high resolution drill core images.

A target of 4~6 dominant minerals were identified in the images based on global and localised colour similarity using Gaussian filters and Watershed segmentation. Statistical analysis of the minerals identified by the Hylogger were ranked, and cross-correlated to the RGB feature ranking.

Results showed good agreement in identifying the dominant mineral types and abundance ranking. In addition, the image processing program was able to identify localised geological features which Hylogger line scanning may overlook.

The program can be applied to the drilling and geology industry for more accurate, accelerated, efficient characterisation of drill core sample on its mineralogy.

ADVICE FOR PARTICIPANTS

Team Arnut recommend keeping an open mind and spending a substantial amount of time considering all possibilities before deciding on the final approach.

They believe the mentors really are worth their weight in gold. It is helpful to bounce ideas off them, as they know the data back to front, and can provide useful advice and guidance along the way.

“For our case, we had to limit the scope of the results, but we figured it was important to present a solution with clear outcomes rather than a solution that was incomplete. Great feedback we received from the judges after the presentation indicated that one of the reasons we won was because the solution we presented was clear and it was obvious how we had used the data” – Wei Chen.

WHAT’S NEXT?

After being announced as the winners of this event, Team Arnut had positive discussions with the challenge owner. They exchanged details and will follow up with the NSW Government.

Everything going well, they plan to continue the development of this solution with the final goal of delivering an innovative and useful solution to industry.

With regard to attending additional hackathons, they don’t currently have any concrete plans in place, but are interested in attending more events.

The team would love to attend Unearthed San Francisco next month, but unfortunately don’t have the necessary resources to get there.

Unearthed wishes Team Arnut every success in their future endeavours.

For more information about upcoming Unearthed hackathons, visit our event page.