Google Teachable Machine 2.0 – Google brings you closer to Machine Learning


Have you always wanted to go deeper into the subject of machine learning and artificial intelligence, but you were frightened by the high complexity in this area? Google has a solution for you that certainly does not cover all use cases, but makes the topic much more understandable and usable for virtually everyone. In the Google Cloud, Google already has a service that is very easy to use and listens to the name AutoML. But even this model you have to train with pictures, etc. and adapt. Techable Machine Version 2.0 also brings Machine Learning to your desktop and is damn easy to use.

In the end, the usage consists of three different steps. First, you need to give material to the Google neural network (images, specific postures or audio) so that the model can be trained. For example, if you want to intelligently detect certain objects in images, then either select images that contain jars, flowers, or whatever, and upload them, or use your computer's webcam and just keep the objects in front of them. Now you say to the model: "These are pictures of a flower" and so on.

Have you fed your data, followed by a simple but important step. The actual training. Here you click more or less only on a button. Experts can still adjust the iterations, learning rate, etc.

If the model has been successfully fed with data and trained, testing will follow. This is also possible directly within the teachable machine. For example, you can use your webcam to let the AI ​​know what's in front of your lens.

Finally, the trained model can be used in either online or offline projects. If you train a model to recognize hand movements, you can, for example, ban a gesture control in your app and much more. Have a look.

Source link



Please enter your comment!
Please enter your name here