Tutorial: Run inference on an object table by using a feature vector model

This tutorial shows you how to create an object table based on the images from the flowers dataset, and then run inference on that object table using the MobileNet V3 model.

The MobileNet V3 model

The MobileNet V3 model analyzes image files and returns a feature vector array. The feature vector array is a list of numerical elements which describe the characteristics of the images analyzed. Each feature vector describes a multi-dimensional feature space, and provides the coordinates of the image in this space. You can use the feature vector information for an image to further classify the image, for example by using cosine similarity to group similar images.

The MobileNet V3 model input takes a tensor of DType tf.float32 in the shape [-1, 224, 224, 3]. The output is an array of tensors of tf.float32 in the shape[-1, 1024].

Required permissions

• To create the dataset, you need the bigquery.datasets.create permission.

• To create the connection resource, you need the following permissions:

  • bigquery.connections.create
  • bigquery.connections.get

• To grant permissions to the connection's service account, you need the following permission:

  • resourcemanager.projects.setIamPolicy

• To create the object table, you need the following permissions:

  • bigquery.tables.create
  • bigquery.tables.update
  • bigquery.connections.delegate

• To create the bucket, you need the storage.buckets.create permission.

• To upload the dataset and model to Cloud Storage, you need the storage.objects.create and storage.objects.get permissions.

• To load the model into BigQuery ML, you need the following permissions:

  • bigquery.jobs.create
  • bigquery.models.create
  • bigquery.models.getData
  • bigquery.models.updateData

• To run inference, you need the following permissions:

  • bigquery.tables.getData on the object table
  • bigquery.models.getData on the model
  • bigquery.jobs.create

Costs

In this document, you use the following billable components of Google Cloud:

• BigQuery: You incur storage costs for the object table you create in BigQuery.
• BigQuery ML: You incur costs for the model you create and the inference you perform in BigQuery ML.
• Cloud Storage: You incur costs for the objects you store in Cloud Storage.

To generate a cost estimate based on your projected usage, use the pricing calculator.

New Google Cloud users might be eligible for a free trial.

For more information on BigQuery storage pricing, see Storage pricing in the BigQuery documentation.

For more information on BigQuery ML pricing, see BigQuery ML pricing in the BigQuery documentation.

For more information on Cloud Storage pricing, see the Cloud Storage pricing page.

Before you begin

Sign in to your Google Cloud account. If you're new to Google Cloud, create an account to evaluate how our products perform in real-world scenarios. New customers also get $300 in free credits to run, test, and deploy workloads.

In the Google Cloud console, on the project selector page, select or create a Google Cloud project.

Roles required to select or create a project

• Select a project: Selecting a project doesn't require a specific IAM role—you can select any project that you've been granted a role on.
• Create a project: To create a project, you need the Project Creator (roles/resourcemanager.projectCreator), which contains the resourcemanager.projects.create permission. Learn how to grant roles.

Note: If you don't plan to keep the resources that you create in this procedure, create a project instead of selecting an existing project. After you finish these steps, you can delete the project, removing all resources associated with the project.

Go to project selector

Verify that billing is enabled for your Google Cloud project.

Enable the BigQuery and BigQuery Connection API APIs.

Roles required to enable APIs

To enable APIs, you need the Service Usage Admin IAM role (roles/serviceusage.serviceUsageAdmin), which contains the serviceusage.services.enable permission. Learn how to grant roles.

Enable the APIs

In the Google Cloud console, on the project selector page, select or create a Google Cloud project.

Roles required to select or create a project

• Select a project: Selecting a project doesn't require a specific IAM role—you can select any project that you've been granted a role on.
• Create a project: To create a project, you need the Project Creator (roles/resourcemanager.projectCreator), which contains the resourcemanager.projects.create permission. Learn how to grant roles.

Note: If you don't plan to keep the resources that you create in this procedure, create a project instead of selecting an existing project. After you finish these steps, you can delete the project, removing all resources associated with the project.

Go to project selector

Verify that billing is enabled for your Google Cloud project.

Enable the BigQuery and BigQuery Connection API APIs.

Roles required to enable APIs

To enable APIs, you need the Service Usage Admin IAM role (roles/serviceusage.serviceUsageAdmin), which contains the serviceusage.services.enable permission. Learn how to grant roles.

Enable the APIs

Create a reservation

To use an imported model with an object table, you must create a reservation that uses the BigQuery Enterprise or Enterprise Plus edition, and then create a reservation assignment that uses the QUERY job type.

Create a dataset

Create a dataset named mobilenet_inference_test:

SQL

1. Go to the BigQuery page.

   Go to BigQuery

2. In the Editor pane, run the following SQL statement:

   CREATE SCHEMA `PROJECT_ID.mobilenet_inference_test`;

   Replace PROJECT_ID with your project ID.

bq

1. In the Google Cloud console, activate Cloud Shell.

   Activate Cloud Shell

2. Run the bq mk command to create the dataset:

   bq mk --dataset --location=us PROJECT_ID:resnet_inference_test

   Replace PROJECT_ID with your project ID.

Create a connection

Create a connection named lake-connection:

Console

1. Go to the BigQuery page.

   Go to BigQuery

2. In the left pane, click Explorer:

   

   If you don't see the left pane, click Expand left pane to open the pane.

3. In the Explorer pane, click Add data.

   The Add data dialog opens.

4. In the Filter By pane, in the Data Source Type section, select Databases.

   Alternatively, in the Search for data sources field, you can enter Vertex AI.

5. In the Featured data sources section, click Vertex AI.

6. Click the Vertex AI Models: BigQuery Federation solution card.

7. In the Connection type list, select Vertex AI remote models, remote functions, BigLake and Spanner (Cloud Resource).

8. In the Connection ID field, type lake-connection.

9. Click Create connection.`

10. In the Explorer pane, expand your project, click Connections, and select the us.lake-connection connection.

11. In the Connection info pane, copy the value from the Service account id field. You need this information to grant permission to the connection's service account on the Cloud Storage bucket that you create in the next step.

bq

1. In Cloud Shell, run the bq mk command to create the connection:

   bq mk --connection --location=us --connection_type=CLOUD_RESOURCE \
   lake-connection
   

2. Run the bq show command to retrieve information about the connection:

   bq show --connection us.lake-connection
   

3. From the properties column, copy the value of the serviceAccountId property and save it somewhere. You need this information to grant permissions to the connection's service account.

Create a Cloud Storage bucket

1. Create a Cloud Storage bucket.
2. Create two folders in the bucket, one named mobilenet for the model files and one named flowers for the dataset.

Grant permissions to the connection's service account

Console

1. Go to the IAM & Admin page.

   Go to IAM & Admin

2. Click Grant Access.

   The Add principals dialog opens.

3. In the New principals field, enter the service account ID that you copied earlier.

4. In the Select a role field, select Cloud Storage, and then select Storage Object Viewer.

5. Click Save.

gcloud

In Cloud Shell, run the gcloud storage buckets add-iam-policy-binding command:

gcloud storage buckets add-iam-policy-binding gs://BUCKET_NAME \
--member=serviceAccount:MEMBER \
--role=roles/storage.objectViewer

Replace MEMBER with the service account ID that you copied earlier. Replace BUCKET_NAME with the name of the bucket you previously created.

For more information, see Add a principal to a bucket-level policy.

Note: There can be a delay of up to a minute before new permissions take effect.

Upload the dataset to Cloud Storage

Get the dataset files and make them available in Cloud Storage:

1. Download the flowers dataset to your local machine.
2. Unzip the flower_photos.tgz file.
3. Upload the flower_photos folder to the flowers folder in the bucket you previously created.
4. Once the upload has completed, delete the LICENSE.txt file in the flower_photos folder.

Create an object table

Create an object table named sample_images based on the flowers dataset you uploaded:

SQL

1. Go to the BigQuery page.

   Go to BigQuery

2. In the Editor pane, run the following SQL statement:

   CREATE EXTERNAL TABLE mobilenet_inference_test.sample_images
   WITH CONNECTION `us.lake-connection`
   OPTIONS(
     object_metadata = 'SIMPLE',
     uris = ['gs://BUCKET_NAME/flowers/*']);

   Replace BUCKET_NAME with the name of the bucket you previously created.

bq

In Cloud Shell, run the bq mk command to create the connection:

bq mk --table \
--external_table_definition='gs://BUCKET_NAME/flowers/*@us.lake-connection' \
--object_metadata=SIMPLE \
mobilenet_inference_test.sample_images

Replace BUCKET_NAME with the name of the bucket you previously created.

Upload the model to Cloud Storage

Get the model files and make them available in Cloud Storage:

1. Download the MobileNet V3 model to your local machine. This gives you a saved_model.pb file and a variables folder for the model.
2. Upload the saved_model.pb file and the variables folder to the mobilenet folder in the bucket you previously created.

Load the model into BigQuery ML

1. Go to the BigQuery page.

   Go to BigQuery

2. In the Editor pane, run the following SQL statement:

   CREATE MODEL `mobilenet_inference_test.mobilenet`
   OPTIONS(
     model_type = 'TENSORFLOW',
     model_path = 'gs://BUCKET_NAME/mobilenet/*');

   Replace BUCKET_NAME with the name of the bucket you previously created.

Inspect the model

Inspect the uploaded model to see what its input and output fields are:

1. Go to the BigQuery page.

   Go to BigQuery

2. In the left pane, click Explorer:

   

3. In the Explorer pane, expand your project, click Datasets, and then click the mobilenet_inference_test dataset.

4. Go to the Models tab.

5. Click the mobilenet model.

6. In the model pane that opens, click the Schema tab.

7. Look at the Labels section. This identifies the fields that are output by the model. In this case, the field name value is feature_vector.

8. Look at the Features section. This identifies the fields that must be input into the model. You reference them in the SELECT statement for the ML.DECODE_IMAGE function. In this case, the field name value is inputs.

Run inference

Run inference on the sample_images object table using the mobilenet model:

1. Go to the BigQuery page.

   Go to BigQuery

2. In the Editor pane, run the following SQL statement:

   SELECT *
   FROM ML.PREDICT(
     MODEL `mobilenet_inference_test.mobilenet`,
     (SELECT uri, ML.RESIZE_IMAGE(ML.DECODE_IMAGE(data), 224, 224, FALSE) AS inputs
     FROM mobilenet_inference_test.sample_images)
   );

   The results should look similar to the following:

   --------------------------------------------------------------------------------------------------------------
   | feature_vector         | uri                                                        | inputs               |
   —-------------------------------------------------------------------------------------------------------------
   | 0.850297749042511      | gs://mybucket/flowers/dandelion/3844111216_742ea491a0.jpg  | 0.29019609093666077  |
   —-------------------------------------------------------------------------------------------------------------
   | -0.27427938580513      |                                                            | 0.31372550129890442  |
   —-------------------------                                                            ------------------------
   | -0.23189745843410492   |                                                            | 0.039215687662363052 |
   —-------------------------                                                            ------------------------
   | -0.058292809873819351  |                                                            | 0.29985997080802917  |
   —-------------------------------------------------------------------------------------------------------------
   

Clean up

Caution: Deleting a project has the following effects:

• Everything in the project is deleted. If you used an existing project for the tasks in this document, when you delete it, you also delete any other work you've done in the project.
• Custom project IDs are lost. When you created this project, you might have created a custom project ID that you want to use in the future. To preserve the URLs that use the project ID, such as an appspot.com URL, delete selected resources inside the project instead of deleting the whole project.

If you plan to explore multiple architectures, tutorials, or quickstarts, reusing projects can help you avoid exceeding project quota limits.

1. In the Google Cloud console, go to the Manage resources page.

   Go to Manage resources

2. In the project list, select the project that you want to delete, and then click Delete.
3. In the dialog, type the project ID, and then click Shut down to delete the project.