Deploy View

Overview

PerceptiLabs Deploy View allows you to export and deploy your model to different targets.

The view displays the following options to select the model(s) to export/deploy:

1. Model Selection Checkbox: enable this to select the model(s) for export/deployment.

2. Search bar: allows you to filter the list of models by name.

To the right of the model selection screen are the export/deployment targets that you can click:

The following subsections describe these targets:

Export Options

The current export options include:

• TensorFlow: exports to TensorFlow's exported model format or to TensorFlow Lite.
• FastAPI Server: generates a TensorFlow model along with a Python server app with a simple API that you can use for inference on your model.

• PL Package: exports a zipped package containing your PerceptiLabs model that you easily can share and load.

Selecting either of these displays a popup with some or all of the following options:

• Save to: allows you to specify the location to which the exported model files are to be placed.

• Optimize (available for TensorFlow model exports): provides options to compress and/or quantize your exported model(s) during export. Selecting either of these options will export to TensorFlow Lite format.

Deployment Options

Select Gradio to export and deploy your model as a Gradio app.

Using the Exported/Deployed Model

After you complete the export/deployment the model can be used for inference.

See Exporting and Deploying Models for information on how to use your exported/deployed model.