Databricks Mosaic AI Integration

Mosaic AI Vector Search is a vector database that is built into the Databricks Data Intelligence Platform and integrated with its governance and productivity tools, and we’ve made it easy to use Mosaic’s vector search capabilities on your computer vision data directly from FiftyOne!

Follow these simple instructions to get started using Mosaic + FiftyOne.

FiftyOne provides an API to create Mosaic indexes, upload vectors, and run similarity queries, both programmatically in Python and via point-and-click in the App.

Note

Did you know? You can search by natural language using Mosaic similarity indexes!

Basic recipe

The basic workflow to use Mosaic to create a similarity index on your FiftyOne datasets and query your data is as follows:

1. Connect to your databricks workspace and create a vector search endpoint.

2. Load a dataset into FiftyOne

3. Compute embedding vectors for samples or patches in your dataset, or select a model to use to generate embeddings

4. Use the compute_similarity() method to generate a Mosaic similarity index for the samples or object patches in a dataset by setting the parameter backend="mosaic" and specifying a brain_key of your choice

5. Use this Mosaic similarity index to query your data with sort_by_similarity()

6. If desired, delete the index

The example below demonstrates this workflow.

Note

You must have access to a databricks account with vector search enabled and install the Databricks Vector Search Python package to run this example:

pip install databricks-vectorsearch

Note that you need to provide credentials as described in this section to avoid entering them manually each time you interact with your Mosaic index.

First, let’s load a dataset into FiftyOne and compute embeddings for the samples:

import fiftyone as fo
import fiftyone.brain as fob
import fiftyone.zoo as foz

# Step 1: Load your data into FiftyOne
dataset = foz.load_zoo_dataset("quickstart")

# Steps 2 and 3: Compute embeddings and create a similarity index
mosaic_index = fob.compute_similarity(
    dataset,
    brain_key="mosaic_index",
    backend="mosaic",
)

Once the similarity index has been generated, you can query your data in FiftyOne by specifying the brain_key:

# Step 4: Query your data
query = dataset.first().id  # query by sample ID
view = dataset.sort_by_similarity(
    query,
    brain_key="mosaic_index",
    k=10,  # limit to 10 most similar samples
)

# Step 5 (optional): Cleanup

# Delete the Mosaic index
mosaic_index.cleanup()

# Delete run record from FiftyOne
dataset.delete_brain_run("mosaic_index")

Note

Skip to this section for a variety of common Mosaic query patterns.

Setup

To get started with Mosaic AI Vector Search, you need to have access to a Databricks workspace which satisfies the requirements for vector search and create a vector search endpoint. You also need to have a catalog and schema in Databricks where you want to create the vector search index.

Installing the Mosaic AI Vector Search client

In order to use the Mosaic backend, you must also install the Databricks Vector Search Python package

pip install databricks-vectorsearch

Using the Mosaic backend

By default, calling compute_similarity() or sort_by_similarity() will use an sklearn backend.

To use the Mosaic backend, simply set the optional backend parameter of compute_similarity() to "mosaic":

import fiftyone.brain as fob

fob.compute_similarity(..., backend="mosaic", ...)

Alternatively, you can permanently configure FiftyOne to use the Mosaic backend by setting the following environment variable:

export FIFTYONE_BRAIN_DEFAULT_SIMILARITY_BACKEND=mosaic

or by setting the default_similarity_backend parameter of your brain config located at ~/.fiftyone/brain_config.json:

{
    "default_similarity_backend": "mosaic"
}

Authentication

You can provide your credentials in a variety of ways.

Environment variables (recommended)

The recommended way to configure your Databricks credentials is to store them in the environment variables shown below, which are automatically accessed by FiftyOne whenever a connection to Databricks is made.

export FIFTYONE_BRAIN_SIMILARITY_MOSAIC_WORKSPACE_URL=https://<unique-url>.cloud.databricks.com/
export FIFTYONE_BRAIN_SIMILARITY_MOSAIC_PERSONAL_ACCESS_TOKEN=XXXXXXXX
export FIFTYONE_BRAIN_SIMILARITY_MOSAIC_CATALOG_NAME=XXXXXXXX
export FIFTYONE_BRAIN_SIMILARITY_MOSAIC_SCHEMA_NAME=XXXXXXXX
export FIFTYONE_BRAIN_SIMILARITY_MOSAIC_ENDPOINT_NAME=XXXXXXXX

This is only one example of variables that can be used to authenticate an Mosaic client. Find more information here.

FiftyOne Brain config

You can also store your credentials in your brain config located at ~/.fiftyone/brain_config.json:

{
    "similarity_backends": {
        "mosaic": {
            "workspace_url": "https://<unique-url>.cloud.databricks.com/",
            "personal_access_token": "XXXXXXXX",
            "catalog_name": "XXXXXXXX",
            "schema_name": "XXXXXXXX",
            "endpoint_name": "XXXXXXXX"
        }
    }
}

Note that this file will not exist until you create it.

Keyword arguments

You can manually provide credentials as keyword arguments each time you call methods like compute_similarity() that require connections to Databricks:

import fiftyone.brain as fob

mosaic_index = fob.compute_similarity(
    ...
    backend="mosaic",
    brain_key="mosaic_index",
    workspace_url = "https://<unique-url>.cloud.databricks.com/",
    personal_access_token = "XXXXXXXX",
    catalog_name = "XXXXXXXX",
    schema_name = "XXXXXXXX",
    endpoint_name = "XXXXXXXX"
)

Note that, when using this strategy, you must manually provide the credentials when loading an index later via load_brain_results():

mosaic_index = dataset.load_brain_results(
    "mosaic_index",
    workspace_url = "https://<unique-url>.cloud.databricks.com/",
    personal_access_token = "XXXXXXXX",
    catalog_name = "XXXXXXXX",
    schema_name = "XXXXXXXX",
    endpoint_name = "XXXXXXXX"
)

Mosaic config parameters

The Mosaic backend supports the following parameter to customize your similarity queries.

• index_name (None): the name of the Mosaic vector search index to use or create. If not specified, a new unique name is generated automatically

You can specify this parameter via any of the strategies described in the previous section. Here’s an example of a brain config that includes all of the available parameters:

{
    "similarity_backends": {
        "mosaic": {
            "index_name": "your-index"
        }
    }
}

However, typically this parameter is directly passed to compute_similarity() to configure a specific new index:

mosaic_index = fob.compute_similarity(
    ...
    backend="mosaic",
    brain_key="mosaic_index",
    index_name="your-index",
)

Managing brain runs

FiftyOne provides a variety of methods that you can use to manage brain runs.

For example, you can call list_brain_runs() to see the available brain keys on a dataset:

import fiftyone.brain as fob

# List all brain runs
dataset.list_brain_runs()

# Only list similarity runs
dataset.list_brain_runs(type=fob.Similarity)

# Only list specific similarity runs
dataset.list_brain_runs(
    type=fob.Similarity,
    patches_field="ground_truth",
    supports_prompts=True,
)

Or, you can use get_brain_info() to retrieve information about the configuration of a brain run:

info = dataset.get_brain_info(brain_key)
print(info)

Use load_brain_results() to load the SimilarityIndex instance for a brain run.

You can use rename_brain_run() to rename the brain key associated with an existing similarity results run:

dataset.rename_brain_run(brain_key, new_brain_key)

Finally, you can use delete_brain_run() to delete the record of a similarity index computation from your FiftyOne dataset:

dataset.delete_brain_run(brain_key)

Note

Calling delete_brain_run() only deletes the record of the brain run from your FiftyOne dataset; it will not delete any associated Mosaic index, which you can do as follows:

# Delete the Mosaic index
mosaic_index = dataset.load_brain_results(brain_key)
mosaic_index.cleanup()

Examples

This section demonstrates how to perform some common vector search workflows on a FiftyOne dataset using the Mosaic backend.

Note

All of the examples below assume you have configured your Databricks account and credentials as described in this section.

Create a similarity index

In order to create a new Mosaic similarity index, you need to specify either the embeddings or model argument to compute_similarity(). Here are a few possibilities:

import fiftyone as fo
import fiftyone.brain as fob
import fiftyone.zoo as foz

dataset = foz.load_zoo_dataset("quickstart")
model_name = "clip-vit-base32-torch"
model = foz.load_zoo_model(model_name)
brain_key = "mosaic_index"

# Option 1: Compute embeddings on the fly from model name
fob.compute_similarity(
    dataset,
    model=model_name,
    backend="mosaic",
    brain_key=brain_key,
)

# Option 2: Compute embeddings on the fly from model instance
fob.compute_similarity(
    dataset,
    model=model,
    backend="mosaic",
    brain_key=brain_key,
)

# Option 3: Pass pre-computed embeddings as a numpy array
embeddings = dataset.compute_embeddings(model)
fob.compute_similarity(
    dataset,
    embeddings=embeddings,
    backend="mosaic",
    brain_key=brain_key,
)

# Option 4: Pass pre-computed embeddings by field name
dataset.compute_embeddings(model, embeddings_field="embeddings")
fob.compute_similarity(
    dataset,
    embeddings="embeddings",
    backend="mosaic",
    brain_key=brain_key,
)

Create a patch similarity index

You can also create a similarity index for object patches within your dataset by including the patches_field argument to compute_similarity():

import fiftyone as fo
import fiftyone.brain as fob
import fiftyone.zoo as foz

dataset = foz.load_zoo_dataset("quickstart")

fob.compute_similarity(
    dataset,
    patches_field="ground_truth",
    model="clip-vit-base32-torch",
    backend="mosaic",
    brain_key="mosaic_patches",
)

Connect to an existing index

If you have already created a Mosaic index storing the embedding vectors for the samples or patches in your dataset, you can connect to it by passing the index_name to compute_similarity():

import fiftyone as fo
import fiftyone.brain as fob
import fiftyone.zoo as foz

dataset = foz.load_zoo_dataset("quickstart")

fob.compute_similarity(
    dataset,
    model="clip-vit-base32-torch",      # zoo model used (if applicable)
    embeddings=False,                   # don't compute embeddings
    index_name="your-index",            # the existing Mosaic index
    brain_key="mosaic_index",
    backend="mosaic",
)

Add/remove embeddings from an index

You can use add_to_index() and remove_from_index() to add and remove embeddings from an existing Mosaic index, respectively.

These methods can come in handy if you modify your FiftyOne dataset and need to update the Mosaic index to reflect these changes:

import numpy as np

import fiftyone as fo
import fiftyone.brain as fob
import fiftyone.zoo as foz

dataset = foz.load_zoo_dataset("quickstart")

mosaic_index = fob.compute_similarity(
    dataset,
    model="clip-vit-base32-torch",
    brain_key="mosaic_index",
    backend="mosaic",
)
print(mosaic_index.total_index_size)  # 200

view = dataset.take(10)
ids = view.values("id")

# Delete 10 samples from a dataset
dataset.delete_samples(view)

# Delete the corresponding vectors from the index
mosaic_index.remove_from_index(sample_ids=ids)

# Add 20 samples to a dataset
samples = [fo.Sample(filepath="tmp%d.jpg" % i) for i in range(20)]
sample_ids = dataset.add_samples(samples)

# Add corresponding embeddings to the index
embeddings = np.random.rand(20, 512)
mosaic_index.add_to_index(embeddings, sample_ids)

print(mosaic_index.total_index_size)  # 210

Retrieve embeddings from an index

You can use get_embeddings() to retrieve embeddings from a Mosaic index by ID:

import fiftyone as fo
import fiftyone.brain as fob
import fiftyone.zoo as foz

dataset = foz.load_zoo_dataset("quickstart")

mosaic_index = fob.compute_similarity(
    dataset,
    model="clip-vit-base32-torch",
    brain_key="mosaic_index",
    backend="mosaic",
)

# Retrieve embeddings for the entire dataset
ids = dataset.values("id")
embeddings, sample_ids, _ = mosaic_index.get_embeddings(sample_ids=ids)
print(embeddings.shape)  # (200, 512)
print(sample_ids.shape)  # (200,)

# Retrieve embeddings for a view
ids = dataset.take(10).values("id")
embeddings, sample_ids, _ = mosaic_index.get_embeddings(sample_ids=ids)
print(embeddings.shape)  # (10, 512)
print(sample_ids.shape)  # (10,)

Querying a Mosaic index

You can query a Mosaic index by appending a sort_by_similarity() stage to any dataset or view. The query can be any of the following:

• An ID (sample or patch)

• A query vector of same dimension as the index

• A list of IDs (samples or patches)

• A text prompt (if supported by the model)

import numpy as np

import fiftyone as fo
import fiftyone.brain as fob
import fiftyone.zoo as foz

dataset = foz.load_zoo_dataset("quickstart")

fob.compute_similarity(
    dataset,
    model="clip-vit-base32-torch",
    brain_key="mosaic_index",
    backend="mosaic",
)

# Query by vector
query = np.random.rand(512)  # matches the dimension of CLIP embeddings
view = dataset.sort_by_similarity(query, k=10, brain_key="mosaic_index")

# Query by sample ID
query = dataset.first().id
view = dataset.sort_by_similarity(query, k=10, brain_key="mosaic_index")

# Query by a list of IDs
query = [dataset.first().id, dataset.last().id]
view = dataset.sort_by_similarity(query, k=10, brain_key="mosaic_index")

# Query by text prompt
query = "a photo of a dog"
view = dataset.sort_by_similarity(query, k=10, brain_key="mosaic_index")

Note

Performing a similarity search on a DatasetView will only return results from the view; if the view contains samples that were not included in the index, they will never be included in the result.

This means that you can index an entire Dataset once and then perform searches on subsets of the dataset by constructing views that contain the images of interest.

Accessing the Mosaic client

You can use the client property of a Mosaic index to directly access the underlying Mosaic client instance and use its methods as desired:

import fiftyone as fo
import fiftyone.brain as fob
import fiftyone.zoo as foz

dataset = foz.load_zoo_dataset("quickstart")

mosaic_index = fob.compute_similarity(
    dataset,
    model="clip-vit-base32-torch",
    brain_key="mosaic_index",
    backend="mosaic",
)

mosaic_client = mosaic_index.client
print(mosaic_client)

Advanced usage

As previously mentioned, you can customize your Mosaic indexes by providing optional parameters to compute_similarity().

Here’s an example of creating a similarity index backed by a customized Mosaic index. Just for fun, we’ll specify a custom index name and populate the index for only a subset of our dataset:

import fiftyone as fo
import fiftyone.brain as fob
import fiftyone.zoo as foz

dataset = foz.load_zoo_dataset("quickstart")

# Create a custom Mosaic index
mosaic_index = fob.compute_similarity(
    dataset,
    model="clip-vit-base32-torch",
    embeddings=False,  # we'll add embeddings below
    brain_key="mosaic_index",
    backend="mosaic",
    index_name="custom-quickstart-index",
)

# Add embeddings for a subset of the dataset
view = dataset.take(10)
embeddings, sample_ids, _ = mosaic_index.compute_embeddings(view)
mosaic_index.add_to_index(embeddings, sample_ids)