Case studies

The East Gippsland Shire will conduct a trial of Magpeye during November 2021 to assess the accuracy of its road condition assessments.

The trial will evaluate the accuracy Magpeye’s perceptual sensory system by comparing it to the results of a human default detection system. 

East Gippsland Shire is responsible for maintaining 2,898km of formed roads. The Network is an essential transport route for tourism, forestry and other major Australian industries. Frontline Data Systems will assist East Gippsland Shire in making an assessment of whether Magpeye can support East Gippsland Shire in providing the best road health possible for it 47,725 residents.

 

The Institute of Public Works Engineering Australasia & Local Government NSW commissioned the Automated Detection of Road Defects Evaluation to “identify a range of new and emerging opportunities in supporting developments for artificial intelligence to identify road defects,” in July 2021.

The capacity of Frontline Data System, & seven other automated road defect detection systems were tested across 295.8km of urban, rural and unsealed roads in Blayney Shire, Canterbury Bankstown, Central Coast, Georges River council.

The report’s conclusions were supportive of both Frontline Data’s System’s ability to accurately & efficiently detect road defects and the transition from manual to automated road defect detection systems more broadly, finding: 

“Frontline Data Systems offers a tailored model that can detect most ordinary road surface defects at human-level accuracy and can be trained to work effectively on dashcam footage of any acceptable nature.”

“It [Frontline Data Systems] was able to detect cracking in 88% of its surveyed sections where another system also identified a crack, indicating strong potential for crack detection.”

“The project has proven the value of artificial intelligence for asset inventory, asset condition assessments and the establishment of maintenance programs for roads and transport assets. The outcomes are applicable to any movements and place assets, including roads, footpaths, parks and other outdoor spaces.”

VicRoads conducted a pilot program with Frontline Data Systems “to establish the viability of Artificial Intelligence in conducting road surface assessments” in 2021 and found that: 

“Whenever the human surveyor detected a defect on a SIR segment (see appendix A) the AI surveyor would detect that same defect on the same segment 75% of the time.”

“In some cases, the AI surveyor detected defects that the human surveyor did not. Visual analysis suggests that many of these defects were identified correctly.”