BTA2024 Award Winning Products Announced! Second round of judging underway!

Completed in 10 minutes! Let's enjoy easy AI development with parents and children using "Teachable Machine"! |With IT enthusiast mom Ogamari's baby tech!

Hello! I am Ogamari.

It is now August, and I hope everyone is enjoying the summer with fireworks displays, festivals, swimming pools, etc.

On days when it is too hot to go outside...we recommend playing on the computer at home.

In this issue, we introduce Teachable Machine, an AI development tool that can be easily enjoyed by parents and children!

What is Teachable Machine?

Teachable Machine" is a service provided by Google that allows you to easily handle machine learning. All you need is a PC with a webcam and an Internet environment! Anyone can easily develop AI, even if they do not know how to program.

This is an AI model I created, which can determine whether the character reflected in the camera is "Sinapshu" or "Cinnamoroll". I took images of both Sinapshu and Cinnamoroll, and made the AI memorize their characteristics to determine which is which. This was also completed in just a few minutes!

By the way, my face was 38% for the Sinapshu element and 62% for the Cinnamoroll element (laughs).

Let's use AI to identify rock-paper-scissors moves!

From here, we would like to create a "rock-paper-scissors move discrimination system" together using the Teachable Machine!

If you make a par toward the camera, the AI will identify "the hand you are making is a par".

First, visit the Teachable Machine website.

https://teachablemachine.withgoogle.com/

This time, since we will be identifying hand movements, select "Pose Project.

A screen like this will appear.

Rename "Class1" to "Goo.

Rename Class2 to "Choki" and classes added with "Add Class" to "Par" respectively.

For each class, register a photo of the actual pose. While continuing to click "long press to record," take a picture of each of the Guchokipa hand shapes.

This time I took about 200-300 pictures of each Guchokipa. The key is to take pictures while moving your wrist so that you can take pictures from various angles!

Once a picture of each Guchokipa has been taken, click on "Train Model".

Click to begin training to create a discrimination rule.

The AI will learn the characteristics of each pose by learning the photos you have taken. Please be careful not to switch browser tabs during training.

After the training, the JAKEN hand identification system is complete! Ka, it's too easy..!

If you actually show the camera a rock-paper-scissors hand, the AI will determine what that hand is.

When I gave the goo, it became "goo 100%"!

I tried a subtle move and the AI had a bit of trouble with it (laughs).

Thus, the Teachable Machine web service can create an AI-based system in just 10 minutes.

If we apply this, we can make people learn the faces of dogs and cats and ask them, "Are you dog-faced or cat-faced? Cat face?" or "Do you have a straight back? or "Do you have a straight back or a hunched back? or "Do you have a straight back or a hunched back?

You might even be able to use it for free research during summer vacation...? Please enjoy it with your family!

Child of the Week.

This was my daughter's first time to go to a fireworks display. She was not startled by the loud noise and started sleeping easily 30 minutes after the show started...!

Ogamari (IT enthusiast)

She is a freelance web creator, computer instructor, and programming learner support, etc. She is a new mother who gave birth in May 2010.
I'll do anything I can do with a computer! I do all kinds of work with my MacBook, which is my partner. My hobbies are watching comedy and bowling (I used to be a bowler, but I am taking a break after having a baby).
https://ogamari.me/