A new app uses Artificial Intelligence (AI) to tell you what your pet is thinking.
It’s called Happy Pets and it analyses an animal’s facial features, tells you what breed it is and indicates which of the five most common animal emotions — happy, angry, neutral*, sad and scared — it is feeling.
The free app, which launched this week, works on Apple iOS or Google Android.
It was developed by a team including Professor Uwe Aickelin, Yunjie Jia, Pei-Yun Sun and Rio Susanto who work in the School of Computing and Information Systems at the University of Melbourne, Victoria.
The Happy Pets app uses AI, or machine learning. It’s technology that’s used for vision in robotics and self-driving cars and it works by taking in images and assigning* importance to them.
One common use for AI is in facial recognition technology. Many of us can unlock our mobile phones with an image of our face and it’s an important part of security processes in, for instance, airports.
But this is the first time this kind of technology has been used for animals.
The researchers had to teach the technology to recognise that an image was an animal — instead of, for instance, a blueberry muffin.
It then had to learn to recognise facial features. This is complicated because photos can be taken in so many different ways: from the side, above, below, in bright light. And different animals can have such different facial features. Think about how different the snouts of a border collie and a bulldog are, for instance.
Once the AI behind Happy Pets had learned all this, it then had to be able to detect emotions based on specific facial features that are associated with each emotion, which it has learnt from thousands of examples. For instance, if a dog tightens its eyes and mouth while changing the position of its ears in a particular way, it’s a sign of being scared.
Scientists know that human faces tell us a lot about what someone is feeling or thinking — that facial expressions are an important part of human communication.
They also know that some emotions (like happiness, fear or surprise) seem to have biologically ‘hardwired’ human expressions that are consistent across ethnicities* and societies, rather than being learnt.
How accurate are the results?
In an online article for Melbourne University’s Pursuit, the researchers write they’re happy with how the app performs.
“Well, we think they are pretty solid (results), having extensively tested the app, but you should judge for yourself,” the researchers wrote.
“At the moment, we only have a limited number of breeds available, so if yours isn’t there, it might get approximated* to the nearest one.”
TESTING THE APP
The Kids News team tested Happy Pets.
We used both photos from the internet and those we took of animals we know.
As expected, it didn’t recognise a photo we took of a lamb — it’s not designed to recognise sheep, only cats and dogs.
We took a photo of a border collie dog we know very well and we were pretty sure was happy at the time. The app was 100 per cent certain it was a border collie and 71 per cent certain it was happy.
We also tried a photo of a cat from the internet and the app correctly identified that it was a cat and the breed (Egyptian Mau) and was 75 per cent certain it was scared. As we don’t know the cat, we can’t know if this was accurate.
The researchers would love feedback about how they could develop the app further. There are contact details in the About section of the app.
GLOSSARY
- neutral: neither one nor the other: good nor bad, fast nor slow, happy nor sad
- assigning: allocating to a particular group
- ethnicities: groups of people that have a common national or cultural tradition
- approximated: was similar to something else
EXTRA READING
Body language study translates dog talk
Posties using whistles to keep dogs away
Siba the standard poodle named best in show
QUICK QUIZ
- Name the five common animal emotions.
- What is AI?
- Describe how a dog’s face could look if it is scared?
- What might happen if the app hasn’t yet learnt about your breed of dog?
- What dog breed did Kids News use to test the app?
LISTEN TO THIS STORY
CLASSROOM ACTIVITIES
1. Pet’s Point of View
Imagine that you are a pet dog (or cat, or sheep!). Write a story for Pets News. Your headline is: ‘Is Happy Pet a Good Thing for Us?’
Time: allow 30 minutes to complete this activity.
Curriculum Links: English, Science
2. Extension
What other features do you think the app developers could include to make Happy Pets even better and more helpful for pet owners?
Time: allow at least 20 minutes to complete this activity
Curriculum Links: English, Science, Design and Technologies
VCOP ACTIVITY
Aside from this, there is also this!
Brackets are a great literacy tool for adding aside comments, or comments that could be covered over and the sentence still makes sense. What’s inside the brackets is extra information.
They can be used for a variety of effects: to add more detail, to add humour, to connect with the reader etc.
My little brother, (the funniest kid I know) got himself into big trouble today.
Select 3 sentences from the article to add an aside comment to using brackets. Think about not only what you want to add to the sentence, but also what effect you are trying to create.
HAVE YOUR SAY: Without an app, describe how you judge an animal’s emotions.
No one-word answers. Use full sentences to explain your thinking. No comments will be published until approved by editors.