Image-aware artificial intelligence model using Teachable Machine
The Teachable Machine, introduced by Google, is a quick and easy model for people who are interested in artificial intelligence but have no expertise in it. You don’t need a coding. It’s simple to use on a web basis, so it also doesn’t cost any computing resources. Following the 2017 version, Teachable Machine 2.0 was released in November 2019. You can learn not only images but also sounds and poses. First of all, We will introduce the image-based model.
Using the webcam, we will train a model in which artificial intelligence distinguishes people from objects and dolls.
After showing three different images, simply press the training button to finish. If you click the Advanced tab under the Train Model button before you start learning, you can actually adjust the hyperparameters that are adjusted during learning, and you can adjust the values such as Epochs, Batch Size, Learning Rate, etc., but it shows excellent performance without any special adjustment.
After finishing the training, you can see exactly what it is on the screen.
As a result,
: Collect and group examples into classes or categories to learn from your computer.
: Train and test the model immediately to ensure that new cases can be classified correctly.
: Export the project model (site, app, etc.). You can download models or host them online for free.
Through this image training, you can create an artificial intelligence model that automatically classifies recycling or a flower-sensitive model.
If Teachable Machine is used in medical fields such as dermatology, it can train artificial intelligence models that doctors accumulate their data and use it for Remote Medical Services. As such, It can be used as a model for students and experts to train and use artificial intelligence models for their own purposes.