A four-wheeled vehicle with Canon imaging gear strapped to it pushes through the undergrowth in far north Queensland. It’s on the hunt. It has no driver and no remote control operation either. Instead, the camera is trained to identify and kill one particular shaped leaf. Once it recognises the targeted flora, herbicide sprayers are released.
This is just one of the projects recently submitted in Canon’s Extreme Imaging competition – this imaging technology is trained to identify the weed Lantana Camara – and it’s serious business. Environmental weeds are aggressive invaders of the Australian landscape. Their impact and removal is costing the Australian grazing sector up to $104 million each year – while also endangering thousands of native Australian species and livestock.
This is an example of machine learning solving real world problems. Machine learning technology provides us with so many opportunities – including the chance to undo some of the damage we’ve inflicted on the planet, as well as gain real business insights never thought possible.
At Howorth Communications, we work on various clients that have machine learning capabilities, and are required to explore ways to communicate the complex offering to a mainstream audience.
When I was first asked to work on a machine learning project, I must admit my eyes glazed over. ‘What does that even mean?’ I thought. More fool me. Instead, I’ve now had the chance to learn about and pitch a number of interesting machine learning stories to media, with a great response. The subject is fascinating and I now jump at the chance to work on any machine learning projects that come my way.
Microsoft’s ‘How old?’ campaign went viral on social media. This was the technology that revealed people’s gender and age when they uploaded a photo of themselves. The application used the data stored in the Azure cloud to learn from thousands of images until it could detect the visual elements that depicted these variables.
Whilst a bit of fun, the campaign sparked serious interest from Manly Council’s CIO, Nathan Rogers. He recognised he could use the same tool to serve a different purpose. By feeding thousands of images from the suburb’s CCTV cameras, the software was able to learn which situations required council ranger attention and send an alert to relevant officers.
Tourist bus waiting bays and disabled parking spaces are monitored 24/7, which in turn saves taxpayers money by preventing unnecessary and constant patrolling. Beyond parking, the machine learning solution is even being considered for crowd control and future community safety measurements, and even to predict the surf conditions.
Founded by three Adelaide University students, Seer Insights is another example of a business using Azure Machine Learning. Wineries face a massive issue in that it is difficult to predict the success of a grape growing season. Their product GrapeBrain is an intelligent system that assists vineyard staff, growers and wineries to improve the accuracy of their yield estimates. By feeding in data from the Bureau of Meteorology, satellite images and soil moisture data, they can predict the quantity and quality of yield. Vineyard owners can then determine the quantity of wine they can produce each season and plan accordingly. Down the track, this has the potential to be used on other crops. As food sources diminish, technology like this could help feed the world.
But as technology continues to streamline business, drive efficiency and increased productivity, the landscape of the workplace is shifting and certain jobs are becoming irrelevant. According to the Committee for Economic Development in Australia, almost 40 per cent of Australian jobs that exist today, have a moderate to high likelihood of disappearing in the next 10 to 15 years due to technological advancements. That said, huge opportunities are also being uncovered every day as machines look at problems in new ways and implement solutions in a fraction of the time. New jobs are being created as more people are needed to make sense of the capabilities and figure out which data sources can drive optimal business impact.
To harness this technology there is now a great need for specialised big data analysts. According to IBM, the world creates an additional 2.5 quintillion bytes of data each year so, to turn this vast quantity of data into useful information, a new workforce is needed. On first consideration this career may seem dry to most. However, after working on machine learning stories and campaigns myself with Canon and Microsoft, I have found the insights fascinating. Translating the intricacies of machine learning into stories that resonate with a mainstream audience has been well-received and the media has a strong appetite for news in this area. Machine learning presents us with opportunities in realms never considered and I’m looking forward to seeing where it will take us as we are only just scratching the surface.