project-preview.png

DogID

Date: 2020
Sharpen.Design prompt: Design an app prototype for a dog park and try to finish in an hour

Scope: This would be a mobile app, designed for people who are actually at a dog park. They might be there with or without dogs. In thinking about what kind of problem to solve, I tried to imagine what kinds of problems/questions I have when I’m at a dog park.

Problem: People who are looking at dogs but might not want to go up to the dogs’ owners might need a way to figure out what kind of dog they’re looking at. Maybe they’re looking to adopt a similar breed of dog, maybe they’re trying to conduct research on this breed, or maybe they’re just curious. This app would tell them what breed of dog this is, through machine-learning.

Assumptions I made: Dog owners wouldn’t find it weird to see folks taking pictures of their dogs. Also, that the app is already mature enough to reach high confidence levels for a large variety of dog breeds.

Loading screen

Loading screen

Imagining the flow from “Oh, I wonder what kind of dog that is” to “Ah, cool! I learned something new about dogs today!” - it started with the person needing to actually take out their phone and open the app. So, I started with a loading screen. Ideally, the app would load quick enough that you would barely see this.

Taking a picture

Taking a picture

The opens up straight into a simple camera interface so that people can quickly take a photo of the dogs in front of them. The only button on the screen, hopefully a clear CTA, would actually take the photo.

Picture taken

Picture taken

One the photo has been taken, the app would indicate what parts of the picture were identified as dogs.

In this case, the app only saw two dog faces. The user should tap on the face they want to learn about.

Dog selected

Dog selected

Once tapped, the app would show its best guess as to what breed this dog is, as well as a percentage confidence and any other guesses it has. The user can tap on any one of those to read about that breed.

The breed information (in this case, the wikipedia article and Google image results) would display underneath the guessed breed. This information would be pulled from Google results.

I chose to display the selected dog in the taken image on this page as well so that the information would be read with the dog in sight. If the user wanted to, they could also easily tap back into the selected image and select another dog to read about.

Help train the AI!

Help train the AI!

If the user scrolls down a bit, they could tap on an option to help the app get smarter. If they knew for sure what the breed was, after tapping this, they would be able to submit a breed name so that the app could add to its understanding of breeds.