PySDK Basics: Running Your First Hosted Inference

Documentation User Guides
1

In this user guide, we will explain how to run inference on the DeGirum AI Hub.

To run inference on an image using PySDK, you need the following:

  • model_name: The name of the machine learning model. In this example, we will use mobilenet_v2_ssd_coco model compiled for DeGirum’s Orca.
  • inference_host_address: Address of the hardware running the ML model.
  • zoo_url: The url of the zoo which contains the model. In this example, we will use DeGirum’s public model zoo.
  • cloud_token: A token to access DeGirum AI Hub. See here for instructions to generate a token.
  • image_source: URL or path or PIL image object or NumPy array

The code below shows a Jupyter Notebook cell that illustrates how to run inference:

import degirum as dg
model = dg.load_model(
    model_name='mobilenet_v2_ssd_coco--300x300_quant_n2x_orca1_1',
    inference_host_address='@cloud',
    zoo_url='degirum/public',
    token=cloud_token,
)
image_source = 'https://raw.githubusercontent.com/DeGirum/PySDKExamples/main/images/TwoCats.jpg' 
res=model(image_source)
res.image_overlay

Running the code should give a result as below:

pysdk_example

Related Examples