Model Hub

The Model Hub screen is where you create and manage your models and is particularly useful for watching the training statuses when you are working with multiple models.

The key elements of the Model Hub are shown here:

1. Import: allows you to specify the location of an existing model to import. See Model Import below.

2. Create: opens the data wizard for you to create a new model. See Model Creation below.

3. Selection: selects all models in the list so that you can perform actions on groups of models (e.g., delete multiple models).

4. Name: lists the names of each model.

5. Training Status: lists the current training status of each model:

   • Waiting: training has not yet started, PerceptiLabs is waiting for the model to be run.

   • Training: the model is training; a progress bar will be shown indicating the current progress.

   • Validation: the model is in the validation phase of the training.

   • Paused: the training is paused.

   • Stop: the training was stopped early.

   • Training Complete: the training completely finished.

6. Duration: the amount of time that the training took to complete.

7. Test History: will be empty if the model hasn't been trained, or will provide the option to test the model if it has been fully trained.

8. Exported: indicates whether the model has been exported or not. Hovering the mouse pointer over this will show you where the file was exported.

9. Last Modified: the date when the model was last saved.

10. Management Buttons: allows you to export your model(s) to your GitHub account, open, or delete model(s).

You can also right click on a model in the list to display options to open, rename, or delete a model:

Right clicking on a model displays actions you can perform on it.

Model Import

You can import an existing model by specifying a local directory or GitHub repo where a model.json file is located. This can be performed by clicking the import button on the Model Hub (shown in the screenshot below) or by selecting File > Import Model and specifying the location.

Note

You will need Git installed before you can import from GitHub.

Model import button the Model Hub screen

Model Creation

Click Create to create a new model:

This displays the New Model popup where you select your data and configure it using PerceptiLabs' Data Wizard. See the Basic Image Recognition Tutorial for information on using the data wizard.

The name you specify for the model will become the name of the directory where the model.json file is stored. The base path defaults to ~/Documents/PerceptiLabs/Default.

Upon completing the popup, PerceptiLabs will create the directory with the model.json file on disk, and the model will be listed on the Model Hub screen. Any time you save the model, PerceptiLabs will update the model's model.json file.