IMGeospatial stuns Anglian Water
In the summer of 2014, Anglian Water’s head of innovation Steve Kaye set out his company’s ambitious aim to “produce a ‘Shop Window’ of innovation to create the water company of the future.”
The plan is to collaborate with innovative companies like IMG to, as Steve says:
“Develop ideas to improve every aspect of our water cycle in the context of the environmental and operational challenges we will face in the future. These ideas may stretch beyond our remit as a water company today, but we will be working with commercial, government and academic partners to implement the best ideas in a major pilot scheme in Newmarket.”
Over the past year, IMG staff have been working hard to develop and implement our Data as a Service solutions for this scheme in a variety of ways, based around the functional capabilities of AIMEE, our Automatic Intelligent Multi-feature Extraction Engine.
Firstly, our Land Use Classification tool helps Anglian understand where potential sources of water pollution from agricultural practices within a Source Protection Zone may be present. Benefits include assisting the company’s compliance with legislation such as the European Water Framework Directive, covering strategies to prevent and control pollution of groundwater.
Then, our Occupancy Probability Index provides Anglian with a capacity prediction dataset incorporating occupancy probabilities derived from the residential building’s attributes such as structure class, height, number of stories etc.
Previously, we’ve spoken about how initial tests on AIMEE’s feature extraction outputs surpassed even our own expectations with an incredible accuracy of over 90 percent across 15 classifications of features, including structures like buildings, walls, roads, curbs and natural features like trees, hedges, lawns and bodies of water.
Well, this week, our Validation SME Dr Marcus Travers finished analysing AIMEE’s outputs for the Newmarket demonstration project and found that the results on more than 15 classifications of feature extraction returned an unprecedented accuracy figure of 95 percent- all achieved automatically. Add to this the ease with which these outputs are scalable up to and including country/continent-wide areas and you’ll begin to understand our confidence in AIMEE.
Steve Kaye shares our enthusiasm too, writing in a letter of recommendation for IMG recently:
“…their solutions have exceeded not only our expectations but also the project brief criteria, redefining what it is possible to achieve in geospatial analysis.”
To say we’re proud of our achievements would be somewhat of an understatement, especially after our team spent last week ‘ground truthing’ AIMEE’s results. That involved trekking around the Suffolk countryside with cameras, maps, GPS gear and datasheets physically confirming that what AIMEE identified as a tree, road, swimming pool or garden shed was, in fact, a tree, a road, a swimming pool or a garden shed…
We’re definitely not resting on our laurels and work is still on-going and we’re busy developing new and innovative ways in which AIMEE can help provide companies with timely, accurate business intelligence. It’s just the beginning of our journey but we feel our small, incredibly talented team are just starting to hit their stride.