Deploy Your Model

Overview

Once you have successfully trained your model, you are ready to deploy it for testing and generating predictions on your data. Deployed models allow you to:
    Use the built-in "Test Model" functionality to upload your own images and visually see the predictions your model generates on new data
    Run inference on the model using the Inference API
    Use the model to automatically annotate dataset images with AutoLabel
To deploy your model, simply start the API from the model version details dashboard.
To access these features, you must be an Admin or Project Manager.

Starting the Inference API

    1.
    Navigate to the model version details of the model version you want to deploy.
    2.
    In the API Status section of the dashboard, the status will display "Inactive". Click "Start API" to deploy the model version.
The API will begin initializing which can take several minutes. When it completes, the status will change to "Active" and the provided endpoint will be ready to accept requests.
Click "Start API" to deploy your model
Initializing the API
Resources are used while the API is active. Please be aware of your usage as it may affect billing. Users should consider stopping the API when not in use.

Model Inferences

Plainsight's Inference API takes an image and returns detections/predictions based on your trained model. Use the provided API endpoint to generate ML predictions with your model version. Read more about how to use the API here:

Stopping the Inference API

When not in use, it's recommended that the API be stopped to prevent unnecessary consumption of resources. This can be done from the model version details of your deployed model.
    1.
    Navigate to the model version details of the model whose deployment you want to stop.
    2.
    In the API Status section of the dashboard, the status will display "Active". Click "Stop API" to stop the deployment and prevent the model from using further resources.
Click "Stop API" to stop a deployed model

Deployment Status

You can monitor the status of the APIs and model deployments by viewing the Model Versions table.
Last modified 2mo ago