Imagine Client for Python

This library allows you to easily connect to doc.ai's Imagine API.

Supported Image Models

API

ImageBasedModel.create(imagine_client, model_id)

Given an ImagineClient (imagine_client) and string (model_id), creates an image model. See the "Supported Image Models" section for a list of supported model ids.

ImageBasedModel.infer(image_file_stream, store=false, metadata=None)

Performs the model's inference on an image. A read stream should be passed as the parameter (image_file_stream).

ImageBasedModel.insert_correction(corrections_dict, metadata=None)

Uploads a correction label to the imagine API, for a given inference.

ImageBasedModel.insert_prediction(predictions_dict, metadata=None)

(FOR EDGE PREDICTIONS) Uploads a prediction label to the imagine API, for a given inference.

Example

from imagine_client_py import ImagineClient, ImageBasedModel

# Required parameters
API_KEY = <VALID_IMAGINE_API_KEY>
IMAGE_PATH = <IMAGE_FILE_PATH>
CLIENT_ID = <ANY_CLIENT_ID>

# Initialize imagine client
imagine_client = ImagineClient({
  'api_key': API_KEY,
  'client_id': CLIENT_ID
})

# Create the 'happy-face' model interface
model = ImageBasedModel.create(imagine_client, 'happy-face')

# Create a read stream to a image file
with open(IMAGE_PATH, 'rb') as image_face_file:
  # Perform the model's inference, on the image.
  result = model.infer(image_face_file)

##Tests

To run tests:

export IMAGINE_API_KEY=<API key> IMAGINE_CLIENT_ID=test
pytest