Tensorflow ml5js

Stable version 1.0.2 (Compatible with OutSystems 11)
Published on 18 March 2020 by 
 (0 ratings)

Tensorflow ml5js


1. To create a model of machine learning go to Teachable Machine

2. In the previous link, you’ll see a different type of Project. In our case, we will be using the Image Project

3. Create at least two classes. You can upload multiple sample images or use the webcam to do a movie and get multiple samples images more quickly.

4. Change the names of the classes to be more readable.

5. When you have, at least, two classes click on the button Train Model.

6. Once the training is complete, you can do a quick test before export the project.

     Click Export Mode. On the popup, select the tab Tensorflow.js and the option Download. Click on "Download my model".

7. Extract the zip file and add the generated files as resources on the e-space. Set the Deploy Action as Deploy to Target Directory.

8. Add the dependency Tensorflow_ml5js to the e-space

    Create a button and with a screen action where you need to call the InitModel client action from the Tensorflow_ml5js.

9. Set the required parameter of the InitModel with the URL of the resource model.json.

10. Create another screen action where you will pass a binary as an input parameter.

       On this screen action, add the client action CallClassifier from the dependency Tensorflow_ml5js.

       Set the required parameter of the CallCassifier with the binary input parameter