Image search

Image search uses an image as a search input to perform vector similarity search.

Additional information

Configure image search

To use images as search inputs, configure an image vectorizer module for your collection.

For details, see the modules reference page:

• img2vec-neural
• multi2vec-clip
• multi2vec-bind

Named vectors

Added in v1.24

Any vector-based search on collections with named vectors configured must include a target vector name in the query. This allows Weaviate to find the correct vector to compare with the query vector.

Python (v4)

from weaviate.classes.query import MetadataQuery

reviews = client.collections.get("WineReviewNV")
response = reviews.query.near_text(
    query="a sweet German white wine",
    limit=2,
    target_vector="title_country",  # Specify the target vector for named vector collections
    return_metadata=MetadataQuery(distance=True)
)

for o in response.objects:
    print(o.properties)
    print(o.metadata.distance)

Python (v3)

# Unfortunately, named vectors are not suppored in the v3 API / Python client.
# Please upgrade to the v4 API / Python client to use named vectors.

JS/TS (Beta)

let result 
const myCollection = client.collections.get('WineReviewNV');

result = await myCollection.query.nearText(['a sweet German white wine'],{
  limit: 2,
  targetVector: 'title_country',
  returnMetadata: ['distance']
})

for (let object of result.objects) {
  console.log(JSON.stringify(object.properties, null, 2));
  console.log(JSON.stringify(object.metadata?.distance, null, 2));
}

JS/TS

result = await client.graphql
  .get()
  .withClassName('WineReviewNV')
  .withNearText({
    concepts: ['a sweet German white wine'],
    targetVectors: ['title_country'],
  })
  .withLimit(2)
  .withFields('title review_body country')
  .do();

console.log(JSON.stringify(result, null, 2));

GraphQL

{
  Get {
    WineReviewNV(
      limit: 2
      nearText: {
        targetVectors: ["title_country"]
        concepts: ["a sweet German white wine"]
      }
    ) {
      title
      review_body
      country
    }
  }
}

By local image path

Use the Near Image operator to execute image search.
If your query image is stored in a file, you can use the client library to search by its filename.

Python (v4)

from pathlib import Path

dogs = client.collections.get("Dog")
response = dogs.query.near_image(
    near_image=Path("./images/search-image.jpg"),  # Provide a `Path` object
    return_properties=["breed"],
    limit=1
)

print(response.objects[0])

client.close()

Python (v3)

response = (
    client.query
    .get("Dog", "breed")
    .with_near_image({"image": "image.jpg"})  # default `encode=True` reads & encodes the file
    .with_limit(1)
    .do()
)

JS/TS (Beta)

Not available yet. Vote for the feature request. DYI code below.

// Read the file into a base-64 encoded string
const contentsBase64 = await fs.promises.readFile('image.jpg', { encoding: 'base64' });
// Query based on base64-encoded image
const myCollection = client.collections.get('Dog');

const result = await myCollection.query.nearImage(contentsBase64,{
  returnProperties: ['breed'],
  limit: 1,
})

console.log(JSON.stringify(result.objects, null, 2));

JS/TS

Not available yet. Vote for the feature request. DYI code below.

// Read the file into a base-64 encoded string
const contentsBase64 = await fs.promises.readFile('image.jpg', { encoding: 'base64' });

// Query based on base64-encoded image
result = await client.graphql
  .get()
  .withClassName('Dog')
  .withNearImage({
    image: contentsBase64,
  })
  .withLimit(1)
  .withFields('breed')
  .do();

console.log(JSON.stringify(result, null, 2));

Example response

{
  "data": {
    "Get": {
      "Dog": [
        {
          "breed": "Corgi"
        }
      ]
    }
  }
}

client.close()

By the base64 representation

You can search by a base64 representation of an image:

Python (v4)

base64_string="SOME_BASE_64_REPRESENTATION"

# Get the collection containing images
dogs = client.collections.get("Dog")

# Perform query
response = dogs.query.near_image(
    near_image=base64_string,
    return_properties=["breed"],
    limit=1
)

print(response.objects[0])

client.close()

Python (v3)

base64_string="SOME_BASE_64_REPRESENTATION"

# Perform query
response = (
    client.query
    .get("Dog", "breed")
    .with_near_image(
        {"image": base64_string},
        encode=False  # False because the image is already base64-encoded
    )
    .with_limit(1)
    .do()
)

print(json.dumps(response, indent=2))

JS/TS (Beta)

const base64String = 'SOME_BASE_64_REPRESENTATION';

// Perform query
const myCollection = client.collections.get('Dog');

const result = await myCollection.query.nearImage(base64String,{
  returnProperties: ['breed'],
  limit: 1,
})

console.log(JSON.stringify(result.objects, null, 2));

JS/TS

const base64String = 'SOME_BASE_64_REPRESENTATION';

// Perform query
let result = await client.graphql
  .get()
  .withClassName('Dog')
  .withNearImage({
    image: base64String,
  })
  .withLimit(1)
  .withFields('breed')
  .do();

console.log(JSON.stringify(result, null, 2));

Example response

{
  "data": {
    "Get": {
      "Dog": [
        {
          "breed": "Corgi"
        }
      ]
    }
  }
}

client.close()

Create a base64 representation of an online image.

You can create a base64 representation of an online image, and use it as input for similarity search as shown above.

Python

import base64, requests

def url_to_base64(url):
    image_response = requests.get(url)
    content = image_response.content
    return base64.b64encode(content).decode("utf-8")

base64_img = url_to_base64("https://upload.wikimedia.org/wikipedia/commons/thumb/1/14/Deutsches_Museum_Portrait_4.jpg/500px-Deutsches_Museum_Portrait_4.jpg")

client.close()

JS/TS

import { readFileSync } from 'fs'

const urlToBase64 = async (imageUrl) => {
  const response = await fetch(imageUrl);
  const content = await response.buffer();
  return content.toString('base64');
}

urlToBase64('https://upload.wikimedia.org/wikipedia/commons/thumb/1/14/Deutsches_Museum_Portrait_4.jpg/500px-Deutsches_Museum_Portrait_4.jpg')
  .then(base64 => { console.log(base64) });

Combination with other operators

A Near Image search can be combined with any other operators (like filter, limit, etc.), just as other similarity search operators.

See the similarity search page for more details.

Related pages

• Connect to Weaviate

Questions and feedback

If you have any questions or feedback, let us know in our user forum.

• Technical questions
• Documentation feedback