'Happy Pets', an app that uses AI to read emotions of pets, is born



If you are a pet owner, you may have been wondering 'what your pet thinks.' In an attempt to analyze the thoughts of such pets, Happy Pets, an app that reads the emotions of pets based on facial expression analysis that makes full use of artificial intelligence (AI), was born.

Ever wondered what your pet is thinking? | Pursuit by The University of Melbourne

https://pursuit.unimelb.edu.au/articles/ever-wondered-what-your-pet-is-thinking

Technology-related announcements are being made at the Splendor in the Grass music festival, which is held in Australia every year, but the event in 2020 was postponed due to the new coronavirus. What has been developed to be announced at Splendour in the Grass in 2020 is ' Happy Pets, ' an application that aims to read the emotions of pets using AI.

Happy Pets is distributed as an iOS and Android app.

Happy Pets on the App Store
https://apps.apple.com/au/app/happy-pets/id1515202735



Happy Pets-Apps on Google Play
https://play.google.com/store/apps/details?id=au.edu.unimelb.eresearch.happypets



In the first place, as psychologists and anthropologists think that facial expression is one of the most important aspects of human communication , the face is an important part to convey thoughts and feelings to others. In fact, some emotions such as happiness, fear, and surprise can be identified as uniform facial expressions across ethnic, social, and community boundaries. In other words, human facial expression-based emotional expressions have been cultivated in the process of evolution, not as a result of cultural learning, and are uniform regardless of the environment in which they were raised.

The research team at the University of Melbourne made an attempt to analyze the relationship between facial expressions and emotions in a manner that applies to animals as well. 'Happy Pets' was born as a result. It seems that Happy Pets had two important steps from a technical point of view, and the first was 'Which animal's emotions do you read with face recognition?' This time, we are trying to read emotions from facial expressions for 'pets' kept by humans. The second is to 'identify the main features and patterns that express emotions', which uses convolutional neural networks .



Neural networks use a mechanism called

supervised learning to learn which output (label) corresponds to which input (image). Image recognition using neural networks teaches children the difference between apples and pears in the same way that algorithms are used to adjust the weights and parameters of functions that transform input and output. By continuing to make adjustments until the training data gives the best results, the algorithm adjusts the apple image so that it is correctly recognized as an apple.

The convolutional neural network is optimized for image recognition, and while functioning like a normal neural network, it is possible to extract and identify features from an image using a method called 'convolution'.

Happy Pets uses its convolutional neural network and is learning to use neural networks to recognize pet breeds by using online datasets. In addition, it seems that it is not possible to distinguish between chihuahua and muffin in the stage before learning.



In addition, Happy Pets says that it makes many fine adjustments to the parameters of the neural network to read the emotions of pets. Specifically, Happy Pets seems to have been adjusted to show the pet's most likely feelings out of the five emotions of happiness, anger, neutrality, sadness, and fear.

In addition, Happy Pets uses thousands of images to learn how to associate emotions with specific facial expressions, which is the basis for detecting pet emotions. For example, if a dog characteristically repositions its ears while tightening its eyes and mouth, it's a sign of fear.

“The University of Melbourne research team is proud to say that it's fairly solid and has been able to test the app extensively, but it's up to you to decide if Happy Pets is accurate,” says Happy Pets. I mentioned that I want you to confirm the accuracy by actually using it.



In addition, at the time of article creation, the kind of pets that can read emotions with Happy Pets is limited. Therefore, it seems that other than the corresponding product types, they are classified as the closest product types and their accuracy is low. In addition, it has been clarified that emotions in cats are harder to read than dogs, and in the future it is planned to improve the accuracy by analyzing the emotions of pets, including not only the face but also the body.

Researchers said, 'Overall, pet emotions are difficult to identify, but identifying human emotions is relatively easy. Therefore, whether human emotions can be related to pet emotions, Whether it has a feature or not may be a more exciting field of research in the future.”

in Mobile,   Software,   Science,   Creature, Posted by logu_ii