An engineer (Ben Hamm) made a cat flap??? Cats, rats, AI,… OH MY!
Machine Learning turns out to be a real boon for this Amazon Engineer, Ben Hamm. It is being used widely along with various other concepts of Computer Science.
Mr. Hamm dealt with a difficult situation and guess what he had to deal with? His sweet little cat, Metric. Surprising, right??? Scroll down to read the actual situation.
Metric, a “serial killer” as his owner mentions, has a very bad habit of hunting down rats and other tiny creatures late in the night between 2 AM and 3 AM. Once he hunts down his prey, he brings the half-dead or dead prey to home. This practice of the cat has badly disturbed Ben and his family. Hence, he hooked up the cat flap in his door to an AI-enabled camera (Amazon’s own DeepLens) and an Arduino-powered locking system.
Lets understand this in depth.
So Ben plans to develop a cat lock on the entrance on Metric’s doorway. The lock will be unlocked if and only if the cat isn’t carrying any prey. If he tries to bring in a prey into the house, then the lock doesn’t open until it gets hold off it.
How was the lock modeled?
Mr. Hamm used an Arduino(https://learn.sparkfun.com/tutorials/what-is-an-arduino/all) and IOT concepts to create the automatic door lock.
Detection: How to tell if Metric is a Murderer or is Normal?
The hottest new tech Machine Learning and Image Processing play their role here. Ben bought a camera, Amazon Deep Lens Camera, sole fully for this purpose. He mounts this camera on the top of the cat’s entrance door and for several months, he captures pictures of Metric to understand his normal and killer moments. Around 23,000 pictures were captured and labelled, and these were trained using Machine Learning on an online data training platform, Sage Maker. This trained data is used for detection.
How does it work?
When Metric approaches the entrance, the Amazon DeepLens Camera captures its picture. This picture is analysed using the trained data set and if the output is “Metric is with a prey”, then the door locks and sends the captured image as a text to Ben, which is handled by arduino. Else, the door doesn’t gets locked.
3 MODELS ARE GENERATED:
After analyzing the whole training set, 3 models were created:
1: Is it a cat or not?
2: Is the cat coming or going?
3: Is the cat holding any prey or not?
Last but not the least, Ben thanks his team and mentors without whom this project would not have completed.
Such remarkable works by enthusiastic engineers like Ben Hamm, helps other fellow engineers to create marvellous works. Commendable job!