Yarra Valley Water is Melbourne’s largest retail water utility, providing essential water and sanitation services to more than 1.8 million people. Yarra Valley Water's service area covers most of Melbourne’s northern and eastern suburbs, from Wallan in the north to Warburton in the east, across approximately 4,000 square kilometres.
Yarra Valley Water recently completed an audit of a statistically significant sample of its water meter fleet across its service area to confirm the condition of the water meters and to identify any potential safety issues associated with access. The information collected will allow Yarra Valley Water to plan their resources more effectively for future activities such as meter exchanges.
Yarra Valley Water's trained meter readers used the ORDITAL solution on a smart phone to take classified photographs of water meters. The meter readers were not required to enter any data about the water meter, but simply locate the meter and take a minimum of two photographs of the water meter.
The result of this collection effort was a large photo archive of indexed and catalogued photographs of the water meter fleet sample. The act of capturing the photos was the information capture exercise. Turning that information into structured data utilized ORDITAL's patented crowd sourcing solution.
The ORDITAL mobile app classifies photos of the water meter as they are taken. This project utilized two classifications - a SITE photo that showed where the water meter was located on the property and a PROFILE photo that showed how the water meter was installed. All photos are geolocated and time-stamped for accuracy.
By classifying photographs we are in a position to understand what attributes we are likely to obtain from a photograph of that class. The classified photograph is then processed with jobs in a crowd sourcing network with multiple independent judgments sought to give us a measure of confidence in our attribute. This is an important differentiation between a traditional audit where the auditor enters data in the field without providing evidence of the device, its condition and location.
The example below shows the results from processing the above photographs. We have generated structured data from the photos that tells us whether there is any obstruction in accessing the meter, whether we have clear access to the meter, what sort of surface is below the meter, what colour pipes are connected to the meter (i.e. copper pipes are easier to work on than galvanized pipes), whether the meter is higher or lower than 150mm (6") above the ground, and whether or not the meter is buried.
|OUTDOOR METER NOT IN CONFINED SPACE||100%|
|METER CLEAR OF VEGETATION/BUSHES||100%|
|0 METER(S) VISIBLE IN PHOTO||100%|
|METER IN PIT||100%|