I’ve been talking about machine learning a lot recently - the idea that we can use a bit of computing brain power to look at data, and make predictions, spot patterns and improve the value we can get from data in education. As an example I often talk about how-old.net, that estimates how old you look from simply looking at a picture of you - it’s a good example because it was built very quickly because of the work we’ve been doing to simplify the process of using machine learning services.
So I was excited to see that Manly council had taken a look at how-old.net, and then decided to use our Internet of Things service in their area - for spotting vehicles parked where or when they shouldn’t in reserved bays. They already had an existing network for 100 CCTV cameras, and so they just added a service to spot when buses stay too long in the tourist bus bays, and when people park in disabled bays without a permit. Here’s some of the story:
|“We’ve got Azure Machine Learning deployed in three locations across Manly at the moment,” Mr Rogers said. “One of them is down at the beach across a large bus parking zone. Lots of buses come to Manly to let tourists out to have a look at the beach, which is great, but we just want to make sure there’s equal access to that space for all tour operators.”
The IT team wrote a program that downloads footage from the camera, passes it and uploads it to Azure Machine Learning.
“Azure Machine Learning has been previously trained on 10,000 ‘normal’ or ‘control’ images from the camera, so each time we upload an image it makes a judgement as to whether it’s normal or whether there’s some sort of anomaly. If Azure Machine Learning tells us there’s an anomaly, we have a script on our end that will email the right people so they can immediately go and check out what’s happening.”
The cutting edge technology is also used in mobility bays located beyond foot patrol distance of the CBD. In the future it may also be used to provide a community safety tool with the ability to recognize crowds patterns and trigger public place management protocols and safety evacuation plans.
“Before we could use Azure Machine Learning to watch cameras, the locations were really monitored on an ad hoc basis or during regular patrols,” Mr Rogers said. “Now that we have Azure Machine Learning, we can target patrols when they are needed and those resources can do other duties when they aren’t needed at those locations.”
You can read more about this story here
The media have picked up the story, with GovernmentNews quoting the council’s Chief Information Officer Nathan Rogers:
|“Before we could use Azure Machine Learning to watch cameras, the locations were really monitored on an ad hoc basis or during regular patrols,” he said. “Now that we have Azure Machine Learning, we can target patrols when they are needed and those resources can do other duties when they aren’t needed at those locations.
“We can move to a just in time deployment of resources. We always try to put technology out where the services are actually delivered, rather than focusing on the back office.”
The technology is surprisingly easy and cheap to employ. Mr Rogers estimates it costs the council about $80 a year to use. He added: “We got the whole thing working in under two hours.”
Solving parent parking problems
It got me thinking about scenarios that it could be helpful in schools and universities:
- Many schools have a traffic patrol just to make sure that chaos doesn’t break out at pickup\dropoff times, and they also have networks of CCTV cameras. You could use exactly the same techniques to monitor parking spaces to ensure that cars didn’t block the school driveway, or stop in the wrong place.
- And if you have disabled parking bays, you could automatically monitor to ensure that the right permit is displayed in the windscreen.
- If you don’t let learner drivers on campus, you could setup a system to create an alert when a vehicle is parked in your campus with L-plates on it.
Keeping the Vice Chancellors parking spot clear every minute of the day
The highest value example I could think of - or the one that you might get the most kudos from - would be monitoring the reserved parking spot for the Vice Chancellor or Principal, and creating an alert for the facilities team if somebody else parks in their spot! I reckon that would be a Monday-morning project for somebody from the IT team, and if you’ve already got a CCTV camera covering the area, you could earn yourself an opportunity for promotion for less than $100!
If you are interested in doing one of these things as a summer holiday project (only a few weeks away now, unless you’re in a university, when it’s probably this week) then read on for ideas:
- Read more about machine learning in education
> Making machine learning in education easier for every day users
- Find a Microsoft partner with the Data Analytics competency, and experience in education, and see what ideas they have to help
> Go to Pinpoint, the Microsoft website for finding partner solutions
Sign up for a free Quick Start Internet of Things half-day workshop to explore this and other scenarios
> Internet of Things Quick Start Consultation
- Learn more about Azure Machine Learning yourself, or talk to your analytics team internally to try an idea out
> Machine Learning documentation and tutorials are here
> Azure Internet of Things Suite documentation and overview
- Read the Microsoft Machine Learning blog
> You’ll find it on TechNet here