A vet advised software engineer Estefannie to monitor her rescue cat Teddy’s bathroom habits and make sure there was no plastic stuck in his intestines and she thought it would be a good subject for a YouTube programme.
She penned a python script and set up a camera with a motion sensor that would trigger every time Teddy used his litterbox so she could tell if he was constipated or not.
Since another cat uses the litterbox she had to write code to make sure she could identify the cat making the poo.
"I did a lot of research and found I need to take a whole bunch of pictures. Luckily I already have a picture taking script that can take several pictures a second,” she explains on her YouTube channel.
Estefanni set up a series of infrared lights and gave the camera an infrared filter so it could see in the dark.
“What’s really cool is that you can’t see the infrared lights to the naked eye but the camera can and cats cannot see infrared lights,” she adds.
After successfully installing the infrared rig, Estefanni took over 50,000 photos of Teddy and her other cat using the litterbox.
Estefannie then had to crop all the photos which were fed into a machine-learning script so the computer understands what Teddy looks like.
Unfortunately, the AI could not identify if the cat was doing number one or two so "using the camera timestamp she calculated exactly how long it takes them to poo or pee,” she adds.
She spent a year on the AI project and the cat still lives, although it might need therapy.