AutoML - Deploy Dog Breeds

Objective

We will continue with the dog breeds model you built over lunch. In this notebook, we will:

1. Deploy the model
2. Do a remote prediction from a python backend
3. Do a remote prediction from a JS frontend

Instructions

You will need the Model ID for the model you built over lunch. In pantheon, goto:

1. Artifical Intelligence => Vision
2. From Vision tab, Select AutoML Classification
3. From AutoML Classification tab, select Models
4. Find your model (i.e., Dog Breeds) and verify status is 'Training Successful'

Setup Environment Variables

You will need to setup the environment variables we used in the previous notebook.



In [ ]:

    
# Set your project ID
PROJECT_ID = "[your-project-id]" #@param {type:"string"}
!gcloud config set project $PROJECT_ID

# This is the default, don't change it
COMPUTE_REGION="us-central1"

import sys

# If you are running this notebook in Colab, run this cell and follow the
# instructions to authenticate your GCP account. This provides access to your
# Cloud Storage bucket and lets you submit training jobs and prediction
# requests.

if 'google.colab' in sys.modules:
  from google.colab import auth as google_auth
  google_auth.authenticate_user()

# If you are running this notebook locally, replace the string below with the
# path to your service account key and run this cell to authenticate your GCP
# account.
else:
  %env GOOGLE_APPLICATION_CREDENTIALS your_path_to_credentials.json

Tutorial

Import libraries and define constants



In [ ]:

    
import tensorflow as tf

import numpy as np

# import the Google AutoML client library
from google.cloud import automl_v1beta1 as automl

Create and Configure an AutoML instance



In [ ]:

    
# Create an AutoML client
client = automl.AutoMlClient()

# Derive the full GCP path to the project
project_location = client.location_path(PROJECT_ID, COMPUTE_REGION)

Set the Full Model ID

Use the model ID you found from your GCS console.



In [ ]:

    
# Set your model ID here
model_id="[my-model-id]"

# Get the full path of the model.
model_full_id = client.model_path(PROJECT_ID, COMPUTE_REGION, model_id)

Deploy the Model



In [ ]:

    
# Deploy the model.
response = client.deploy_model(model_full_id)

# Wait for deployment to complete
print(response.result)

Make a local copy of the test image



In [ ]:

    
IMAGE_PATH="gs://" + PROJECT_ID + "-vcm/dog_breeds/fff43b07992508bc822f33d8ffd902ae.jpg"
!gsutil cp $IMAGE_PATH ./

Make an online Prediction



In [ ]:

    
# Specify a path to an Image
IMAGE_PATH="fff43b07992508bc822f33d8ffd902ae.jpg"

# Get the full path of the model.
model_full_id = client.model_path(PROJECT_ID, COMPUTE_REGION, model_id)

# Create client for prediction service.
prediction_client = automl.PredictionServiceClient()

# Read the image and assign to payload.
with open(IMAGE_PATH, 'rb') as image_file:
    content = image_file.read()
payload = {"image": {"image_bytes": content}}

# params is additional domain-specific parameters.
# score_threshold is used to filter the result
# Initialize params
params = {}
score_threshold = None
if score_threshold:
    params = {"score_threshold": score_threshold}

response = prediction_client.predict(model_full_id, payload, params)
print("Prediction results:")
for result in response.payload:
    print("Predicted class name: {}".format(result.display_name))
    print("Predicted class score: {}".format(result.classification.score))

Use from Front-End

We will show using CURL.

TODO:

Why is model_id blank
Get the bytes for the payload



In [ ]:

    
!curl -X POST \
  -H "Authorization: Bearer $(gcloud auth application-default print-access-token)" \
  -H "Content-Type: application/json" \
  https://automl.googleapis.com/v1beta1/projects/${PROJECT_ID}/locations/us-central1/models/${model_id}:predict \
  -d '{ "payload" : { "image": { "imageBytes" : "/9j/4AAQSkZJRgABAQAAAQ … "},}}'

Undeploy the Model



In [ ]:

    
# Undeploy the model.
response = client.deploy_model(model_full_id)

# Wait for undeployment to complete
print(response.result)

Cleaning up

To clean up all GCP resources used in this project, you can delete the GCP project you used for the tutorial.