We thought it might be a great idea to celebrate Pi-Day with a project on a Raspberry Pi. Not just a project, but recognizing cows using OpenCV on a Raspberry Pi!
If you’re not in the know: we are researching to make a comparative list of pattern recognition software of Google, AWS, Microsoft and the like. We used hand made foam board cows to test the software in a consistent way. Read more in our article here.
Pi Day, Raspberry Pi’s and our Cows
Hayo Rubingh thought our pattern recognition research could be a bit more mobile. So he started experimenting with OpenCV, a Open Source pattern recognition library. And he did so on his Raspberry Pi 3, because it’s Pi-Day … Get it?!
*"I never thought this might work, but the [Raspberry] Pi is a fast little computer."*
Recognizing with OpenCV means data, a lot of data
OpenCV works differently than the cloud services we tested before. Rather than learning patterns ‘on the go’, it needs to learn all patterns beforehand.
Hayo used about 120 photo’s of Maxima and Lila (our foam board cows) and 1.700 negative photos (random photo’s without our cows). All those photos build a model.
*“We build our own models this because we wanted to find out how difficult it is. Building your own is something that doesn't seem possible with cloud services yet. But you never know .... ” *
Will Farmers use Raspberry Pi’s
Although this all begun as a lark, because it’s Pi-day we did get more information on how small devices would work with pattern recognition.
We’re not sure if this particular model of Raspberry Pi is the best configuration, mostly because of their limited power.
Strong, mobile computers like your iPhone 7, or Samsung S7 with multiple camera’s might be perfect platforms to get to the next level of real time mobile pattern recognition.
*“I think the future of pattern recognition might be a combination of two worlds: the realtime and multi cam abilities of a mobile device and the vast data depth of our cognitive cloud services.” *
Any way, happy Pi-Day!
This was a great example of how we do research, but most of all to celebrate innovation.
Want to stay posted about AI for your industry?
Want to know more, or stay in touch with our cognitive services research? Follow our blog: blog.mirabeau.nl and if you have specific questions, please get in touch with Edgard Beckand, who leads Mirabeau Labs email@example.com