Nilvana Vision Studio - API Endpoints
Please follow the instructions to deploy your model to get predictions with nilvana vision studio.
The last mile in deep learning application development: Model Serving. Once you have obtained a satisfactory model through the automated model training tool, you can then utilize the API Endpoint tool to package the model, generate model endpoints, and perform object detection through the sample code we've provide. If you are unsure how to perform automated model training, please refer to "Nilvana Vision Studio - Model Training".
Generating API Endpoints
With vision studio's API Endpoint tool, you can simplify the model deployment process by clicking the create button for any specific model API Endpoint or proceed to the API Endpoints page to create your own model endpoint.
There are two options in the Create API Endpoint dialog box that affect the recognition results of the model application, namely Dimensions and Minimum Probability Threshold. The recommended dimensions are the same as the training dimensions used for model training, which will result in better recognition results. The appropriate minimum probability threshold value depends on the application context. The higher the value, the higher the chance of minimizing false positives, but the higher chance of missing objects as well. In addition, it is also possible to select the category to be used in the application context.
Accessing API Endpoint Information
Once an API endpoint has been created, its information can be accessed in the endpoint list. By clicking on the sample code button for any specific endpoint, we provide a variety of programming examples to help you navigate through the endpoints and build your own object detection applications.